Types of sppr. Decision support systems (DSS) general concept of DSS

The purpose of writing this article was to short review principles of building Intelligent Decision Support Systems ( IDSS), the role of machine learning, game theory, classical modeling and examples of their use in DSS. The purpose of the article not is to drill deep into the heavy theory of automata, self-learning machines, as well as BI tools.

Introduction

There are several definitions IDSS, which, in general, revolve around the same functionality. IN general view, IDSS is such a system that assists decision makers (Decision Makers) in making these very decisions, using data mining, modeling and visualization tools, has a friendly (G)UI, stable in quality, interactive and flexible in settings.

Why do we need DSS:

1. Difficulty in making decisions
2. The need for an accurate evaluation of various alternatives
3. The need for predictive functionality
4. The need for multi-threaded input (to make a decision, you need conclusions based on data, expert opinions, known limitations, etc.)

The first DSS (at that time still without I) grew out of TPS (Transaction Processing Systems), in the mid-60s - early 70s. Then these systems did not have any interactivity, representing, in fact, add-ons over the RDBMS, with some (not at all great) functionality numerical simulation. One of the first systems can be called DYNAMO, developed in the depths of MIT and representing a system for simulating any processes based on historical transactions. After the IBM 360 mainframes entered the market, conditional commercial systems began to appear, which were used in the defense industry, special services and research institutes.

Since the early 1980s, we can already talk about the formation DSS subclasses such as MIS (Management Information System), EIS (Executive Information System), GDSS (Group Decision Support Systems), ODSS (Organization Decision Support Systems), etc. In fact, these systems were frameworks capable of working with data on various hierarchy levels (from individual to corporate), and inside it was possible to introduce any kind of logic. An example is the GADS (Gate Assignment Display System) system developed by Texas Instruments for United Airlines, which supported decision making in Field Operations - assigning gates, determining the optimal parking time, etc.

In the late 80s there were PSSPR(Advanced - Advanced), which allowed for "what-if" analysis and used more advanced modeling tools.

Finally, since mid 90s began to appear and IDSS, which were based on the tools of statistics and machine learning, game theory and other complex modeling.

Diversity of DSS

At present, there are several ways classification DSS, we will describe 3 popular ones:

By area of ​​application

• Business and management (pricing, workforce, products, strategy, etc.)
• Engineering (product design, quality control...)
• Finance (lending and loans)
• Medicine (drugs, treatments, diagnostics)
• Environment

By data\model ratio(Stephen Alter method)

• FDS (File Drawer Systems - systems for providing access to the necessary data)
• DAS (Data Analysis Systems - systems for fast data manipulation)
• AIS (Analysis Information Systems - data access systems according to the type of solution required)
• AFM(s) (Accounting & Financial models (systems) - systems for calculating financial consequences)
• RM(s) (Representation models (systems) - simulation systems, AnyLogic as an example)
• OM(s) (Optimization models (systems) - systems that solve optimization problems)
• SM(s) (Suggestion models (systems) - rule-based inference systems)

By type of instrument used

• Model Driven - based on classical models (linear models, inventory management models, transport, financial, etc.)
• Data driven - based on historical data
• Communication Driven - systems based on group decision-making by experts (systems for facilitating the exchange of opinions and calculating average expert values)
• Document Driven - essentially an indexed (often multidimensional) document storage
• Knowledge driven - suddenly, based on knowledge. What does the knowledge of both expert and machine-derived

I want a complaint book! normal DSS

Despite such a variety of classification options, the requirements and attributes of the DSS fit well into 4 segments:

1. Quality
2. Organization
3. Restrictions
4. Model

In the diagram below, we will show exactly which requirements and which segments fall into:

Separately, we note such important attributes as scalability (in the current agile approach, you can’t do without it), the ability to process bad data, usability and user-friendly interface, and undemanding resources.

Architecture and design of IDSS

There are several approaches to how to architecturally represent a DSS. Perhaps the best description of the difference in approaches is "who is into what much." Despite the variety of approaches, attempts are being made to create some kind of unified architecture, at least at the top level.

Indeed, DSS can be divided into 4 large layers:

1. Interface
2. Modeling
3. data mining
4. data collection

And in these layers you can cram any kind of tools.

In the diagram below I present my vision of the architecture, with a description of the functionality and examples of tools:

The architecture is more or less clear, let's move on to the design and actual construction of the DSS.

In principle, there is no rocket science here. When building an IDSS, the following steps must be followed:

1. Domain analysis (actually, where we will use our IDSS)
2. Data collection
3. Data analysis
4. Choice of models
5. Expert analysis\interpretation of models
6. Implementation of models
7. IDSS evaluation
8. Implementation of IDSS
9. Collecting feedback ( at any stage, in fact)

On the diagram it looks like this:

There are two ways to evaluate the IDSS. First, by the attribute matrix, which is presented above. Secondly, according to the criteria checklist, which can be anything and depend on your specific task. As an example of such a checklist, I would give the following:

I emphasize that this is only IMHO and you can make a checklist that is convenient for yourself.

Where is machine learning and game theory?

Yes, almost everywhere! At least in the modeling layer.

On the one hand, there are classic domains, let's call them "heavy", such as supply chain management, production, inventories, and so on. In heavy domains, our favorite algorithms can bring additional insights to established classical models. Example: predictive analytics for equipment failures (machine learning) will work great with some kind of FMEA analysis (classic).

On the other hand, in "light" domains, like customer analytics, churn prediction, loan repayments, machine learning algorithms will be in the foreground. And in scoring, for example, you can combine the classics with NLP when deciding whether to issue a loan based on a package of documents (just the same document driven DSS).

Classical machine learning algorithms

Let's say we have a task: a sales manager for steel products needs to understand at the stage of receiving an application from a client what quality finished products will go to the warehouse and apply some control action if the quality is lower than required.

Let's do it very simply:

Step 0. Determine the target variable (well, for example, the content of titanium oxide in the finished product)
Step 1. Decide on the data (upload from SAP, Access and in general from everywhere we can reach)
Step 2. Collecting features\generating new ones
Step 3. Draw the data flow process and launch it into production
Step 4. Select and train the model, run it on the server
Step 5. Define feature importances
Step 6. Decide on the input of new data. Let our manager enter them, for example, by hand.
Step 7. We write a simple web-based interface on the knee, where the manager enters the values ​​​​of important features with handles, it spins on a server with a model, and the predicted product quality is spit out into the same interface

Voila, kindergarten-level IDSS is ready, you can use it.

Similar "simple" algorithms are also used by IBM in its Tivoli DSS, which allows you to determine the state of your supercomputers (Watson in the first place): based on logs, information on Watson performance is displayed, resource availability, cost vs profit balance, maintenance needs, etc. are predicted.

Company ABB offers its customers DSS800 to analyze the operation of electric motors of the same ABB on a paper line.

Finnish Vaisala, a manufacturer of sensors for the Finnish Ministry of Transport, uses IDSS to predict when de-icing should be applied on roads to avoid accidents.

Again Finnish. Foredata offers IDSS for HR, which helps to make decisions on the suitability of a candidate for a position even at the stage of selecting a resume.

At Dubai Airport, a DSS operates in the cargo terminal, which determines the suspicious nature of the cargo. Under the hood, algorithms, based on accompanying documents and data entered by customs officers, highlight suspicious cargoes: the features are the country of origin, information on the packaging, specific information in the declaration fields, etc.

Thousands of them!

Conventional Neural Networks

In addition to simple ML, Deep Learning fits perfectly into the DSS.

Some examples can be found in the military-industrial complex, for example, in the American TACDSS (Tactical Air Combat Decision Support System). There, neurons and evolutionary algorithms are spinning inside, helping in determining friend or foe, in assessing the probability of hitting a salvo at a given moment, and other tasks.

In a slightly more real world, consider this example: in the B2B segment, you need to determine whether to issue a loan to an organization based on a package of documents. It is in B2C that the operator torments you with questions over the phone, puts down the values ​​of the features in his system and announces the decision of the algorithm, in B2B it is somewhat more complicated.

IDSS can be built there as follows: a potential borrower brings a pre-agreed package of documents to the office (well, or sends scans by email, with signatures and seals, as expected), documents are fed into OCR, then transferred to the NLP algorithm, which further divides the words into features and feeds them to NN. The client is asked to drink coffee (at best), or that's where the card was issued there and go to come after lunch, during which time everything will be calculated and display a green or red smiley on the screen of the operator girl. Well, or yellow, if it seems ok, but the god of information needs more information.

Similar algorithms are also used in the Ministry of Foreign Affairs: a visa application form + other certificates are analyzed directly at the embassy / consulate, after which one of 3 emoticons is displayed on the screen for the employee: green (issue a visa), yellow (have questions), red (applicant in the stop list ). If you have ever received a visa to the USA, then the decision that the consular officer gives you is precisely the result of the algorithm in conjunction with the rules, and not his personal subjective opinion about you :)

In heavy domains, DSSs based on neurons are also known, which determine the places of buffer accumulation on production lines (see, for example, Tsadiras AK, Papadopoulos CT, O'Kelly MEJ (2013) An artificial neural network based decision support system for solving the buffer allocation problem in reliable production lines. Comput Ind Eng 66(4):1150–1162), Min-Max General Fuzzy Neural Networks (GFMMNN) for Water Consumer Clustering ( Arsene CTC, Gabrys B, Al-Dabass D (2012) Decision support system for water distribution systems based on neural networks and graphs theory for leakage detection. Expert Syst Appl 39(18):13214–13224) and others.

In general, it is worth noting that NNs are the best suited for making decisions under uncertainty, i.e. conditions in which the real business lives. Clustering algorithms also fit in well.

Bayesian networks

It sometimes happens that our data is heterogeneous in terms of types of occurrence. Let's take an example from medicine. A patient came to us. We know something about him from the questionnaire (gender, age, weight, height, etc.) and anamnesis (past heart attacks, for example). Let's call this data static. And we learn something about him in the process of periodic examination and treatment (several times a day we measure the temperature, blood composition, etc.). We call this data dynamic. It is clear that a good DSS should be able to take into account all these data and issue recommendations based on the fullness of the information.

Dynamic data is updated in time, respectively, the pattern of the model will be as follows: learning-solution-learning, which in general is similar to the work of a doctor: roughly determine the diagnosis, drip the medicine, look for the reaction. Thus, we are constantly in a state of uncertainty whether the treatment will work or not. And the patient's condition changes dynamically. Those. we need to build a dynamic DSS, and also knowledge driven.

In such cases, Dynamic Bayesian Networks (DBN) - a generalization of models based on Kalman filters and Hidden Markov Models - will help us a lot.

Let's divide the data on the patient into static and dynamic.

If we were building a static Bayesian grid, then our task would be to calculate the following probability:

,

Where is the node of our grid (the top of the graph, in fact), i.e. the value of each variable (sex, age....), and C is the predicted class (illness).

The static grid looks like this:

But it's not ice. The patient's condition is changing, time is running out, it is necessary to decide how to treat him.

This is what DBS is for.

First, on the day of the patient's admission, we build a static grid (as in the picture above). Then, every day i we build a grid based on dynamically changing data:

Accordingly, the aggregate model will take the following form:

Thus, we calculate the result according to the following formula:

Where T- cumulative time of hospitalization, N- the number of variables at each of the DBS steps.

It is necessary to introduce this model into the DSS in a slightly different way - rather, here it is necessary to go from the opposite, first fix this model, and then build an interface around. That is, in fact, we have made a hard model, inside which there are dynamic elements.

Game theory

Game theory, in turn, is much better suited for IDSS, created for strategic decision making. Let's take an example.

Suppose there is an oligopoly on the market (a small number of rivals), there is a certain leader, and this (alas) is not our company. We need to help the management make a decision about the volumes of our products: if we produce products in volume, and our rival -, will we go into the red or not? To simplify, let's take a special case of an oligopoly - a duopoly (2 players). While you are thinking, RandomForest is here or CatBoost, I will suggest you use the classic - Stackelberg equilibrium. In this model, the behavior of firms is described by a dynamic game with complete perfect information, while the feature of the game is the presence of a leading firm, which first sets the volume of output of goods, and the rest of the firms are guided in their calculations by it.
To solve our problem, we just need to calculate such , which will solve the optimization problem of the following form:

To solve it (surprise-surprise!) you just need to equate the first derivative with respect to zero.

At the same time, for such a model, we only need to know the offer on the market and the cost per product from our competitor, then build a model and compare the resulting q with the one that our management wants to throw on the market. Agree, it is somewhat easier and faster than sawing NN.

For such models and DSS based on them, Excel is also suitable. Of course, if the input data needs to be calculated, then something more complicated is needed, but not much. The same Power BI can handle it.

Looking for a winner in the battle of ML vs ToG is pointless. Too different approaches to solving the problem, with their pluses and minuses.

What's next?

FROM state of the art IDSS seems to have figured out where to go next?

In a recent interview, Judah Pearl, the creator of those same Bayesian networks, made an interesting point. To rephrase slightly,

“All machine learning experts are doing right now is fitting a curve to the data. Fitting is not trivial, complicated and dreary, but still fitting. ”

(read)

Most likely, wangyu, in 10 years we will stop hard-coding models, and instead start teaching computers everywhere in the simulated environments created. Probably, the implementation of IDSS will go along this path - along the path of AI and other skynets and WAPRs.

If we look at a closer perspective, then the future of IDSS lies in the flexibility of decisions. None of the proposed methods (classical models, machine learning, DL, game theory) is universal in terms of efficiency for all tasks. A good DSS should combine all these tools + RPA, while different modules should be used under different tasks and have different output interfaces for different users. A sort of cocktail, mixed, but by no means shaken.

Literature

1. Merkert, Mueller, Hubl, A Survey of the Application of Machine Learning in Decision Support Systems, University of Hoffenheim 2015
2. Tariq, Rafi,Intelligent Decision Support Systems - A Framework, India, 2011
3. Sanzhez i Marre, Gibert, Evolution of Decision Support Systems, University of Catalunya, 2012
4. Ltifi, Trabelsi, Ayed, Alimi, Dynamic Decision Support System Based on Bayesian Networks, University of Sfax, National School of Engineers (ENIS), 2012

Decision support systems(DSS) are computer systems, almost always interactive, designed to assist a manager (or supervisor) in making decisions. DSSs include both data and models to help the decision maker solve problems, especially those that are poorly formalized. The data is often retrieved from an online query processing system or database. The model can be a simple "gain and loss" type to calculate profit under some assumptions, or a complex optimization model type to calculate the load for each machine on the shop floor. DSS and many of the systems discussed in the following sections are not always justified by the traditional cost-benefit approach; for these systems, many of the benefits are intangible, such as deeper decision making and better understanding of data.

Rice. Figure 1.4 shows that a decision support system requires three primary components: a management model, a data management to collect and manually process data, and a dialogue management to facilitate user access to the DSS. The user interacts with the DSS through a user interface, selecting the particular model and dataset to use, and then the DSS presents the results to the user through the same user interface. The management model and data management operate largely behind the scenes and range from a relatively simple generic model in a spreadsheet to a complex, complex planning model based on mathematical programming.

Rice. 1.4. Decision Support System Components

An extremely popular type of DSS is in the form of a financial statement generator. Using a spreadsheet such as Lotus 1-2-3 or Microsoft Excel, models are created to predict various elements of an organization or financial condition. The previous financial statements of the organization are used as data. The initial model includes various assumptions about future trends in the expenditure and income categories. After considering the results of the base model, the manager conducts a series of "what if" studies, changing one or more assumptions to determine their impact on the initial state. For example, a manager might probe the impact on profitability if new product sales grew by 10% annually. Or a manager might investigate the impact of a larger than expected increase in the price of raw materials, such as 7% instead of 4% annually. This type of financial statement generator is a simple yet powerful DSS to guide financial decision making.

An example of a DSS to bring data transactions into effect is the police outreach budgeting system used by cities in California. This system allows the police officer to see the map and displays the geographic area data, shows the police call tones, call types and call times. The system's interactive graphics capability allows the officer to manipulate the map, zone, and data to quickly and easily suggest variations of police exit alternatives.

Another example of DSS is an interactive system for volume planning and production in a large paper company. This system uses detailed previous data, predictive and planning models to play on the computer general indicators companies under different planning assumptions. Most oil companies are developing DSS to support capital investment decision making. This system includes various financial conditions and models for creating future plans, which can be presented in tabular or graphical form.

All examples of DSSs given are referred to as specific DSSs. They are the actual applications that help in the decision making process. In contrast, a decision support system generator is a system that provides a set of capabilities to quickly and easily build specific DSSs. The DSS Generator is a software package designed to run on a partially computer basis. In our financial statement example, Microsoft Excel or Lotus 1-2-3 can be considered DSS generators, while Excel or Lotus 1-2-3 models for designing financial statements for a private branch of a company are specific DSSs.

DSS is discussed in more detail in Sect. 2.2.

DSS appeared mainly due to the efforts of American scientists in the late 1970s and early 1980s, which was largely facilitated by the widespread use of personal computers, standard application software packages, as well as significant advances in the creation of artificial intelligence (AI) systems.

Distinctive features of the SPPR.

DSS is characterized by the following distinctive features.

Orientation towards solving poorly structured (formalized) tasks, typical mainly for high levels of management;

Possibility of combination traditional methods access and processing of computer data with the capabilities of mathematical models and methods for solving problems based on them;

Orientation to the non-professional end user of the computer through the use of an interactive mode of operation;

High adaptability, providing the ability to adapt to the features of the available hardware and software, as well as user requirements.

Place of DSS among information systems. The information model of an organization can be thought of as the following hierarchical model, which includes the following three levels (see Figure 4.3):

Data processing,

Data processing,

Making decisions.

Rice. 4.3. Hierarchy of information systems in the company

At the first lowest level are SEOD. In the hierarchy of management decisions, this level corresponds to the level of management control that automates the workflow in the organization. The main characteristics of the SOED are:

Data processing at the operational control level,

Efficient processing of commercial transactions conducted by the organization,

Scheduling and optimizing computer performance,

Integration of files describing related tasks,

Compiling reports for management.

At the second middle level, corresponding to the level of managerial control, the emphasis is shifted to the information processing procedures performed by the MIS. This processing usually refers to the planning of activities in such functional areas of the organization as marketing, production, finance, accounting, personnel. The main characteristics of the IMS should be considered:

Preparing information useful at the level of middle management,

Structuring (ordering) information flows,

Integration (combination) of data received from the SEOD in the functional areas of business (IMS marketing, MIS production, etc.),

Creating a query-response system and reporting to management (usually using databases).

At the third highest level of management, corresponding to strategic planning, the most important decisions of the organization are formed. DSS used at this level (as will be clear from what follows, DSS can be used at any level of management) have the following characteristics:

Preparation of solutions for senior management,

Ensuring high adaptability to changes and high speed responses to user requests,

Providing decision-making assistance to any individual managers.

Data management in the SEDI environment is carried out mainly to process the current business operations conducted by the firm. The creation of the IMS was associated with the advent of the DBMS, which made it possible to organize the modes of queries, data processing, and the creation of various management reports. However, the main advantage of creating a DBMS was to reduce the cost of ongoing programming associated with the operation of databases. It should be pointed out that the requirements imposed by the user on such systems are relatively low. The requirements for DSS are much more serious. This concerns the growing need for reliable data, including those of a probabilistic nature, as well as the tightening of time restrictions on the query mode and the use of data coming from non-computerized sources. Compliance with such requirements ensures rapid data exchange between the databases included in the DSS and a large database that stores information about the firm's operations.

So, SEOD and MIS make it possible to satisfy the information needs of the user through quick access to the necessary data and receiving reports (built with varying degrees of data processing) that facilitate decision making. In the case of DSS, it is more correct to speak about the ability of the system, together with the user, to create new information(often in the form of ready-made alternatives) for decision making.

It should be noted that the considered approach to establishing the place of DSS among IS may be somewhat misleading for the reader. Thus, it may seem that DSS can only be used at the highest levels of government. In fact, they can be used to help decision making at any level of management. In addition, decisions made at different levels of government often need to be coordinated. Therefore, an important function of the DSS is the coordination of decision makers at different levels of government, as well as within the same level. And, finally, it may seem to the reader that decision-making assistance is the only thing that higher-level management may need from information systems. However, decision making is only one of the functions of managers for which they receive assistance from information systems.

Note also that the term "management information systems" itself is used in the literature in a broad and narrow sense. In a broad sense, it includes any types of considered computer systems (SEOD, MIS, DSS, etc.) used in the interests of managers. In a narrow sense, this term means a type of IS that produces management reports, i.e. ISU.

Structure of the DSS

Until now, we have not touched upon the structure of the DSS, considering it to be some kind of "black box". The first idea about the structure of the DSS can be drawn from the consideration of Fig. 4.4.

In addition to the user, the DSS includes three main components: a subsystem for processing and storing data, a subsystem for storing and using models, and a software subsystem. The latter includes a database management system (DBMS), a model database management system (BMS) and a user-computer dialogue management system (UDC).

data subsystem. The data processing and storage subsystem is characterized by all the known advantages of building and using databases. However, the use of databases as part of the DSS is characterized by certain features (see Fig. 4.5). For example,

Rice. 4.4. Structure of the DSS

DSS databases have a much larger set of data sources, including external sources that are especially important for decision-making at high levels of management, as well as sources of non-computerized data. Another feature is the possibility of preliminary "compression" of data coming from multiple sources, through their preliminary joint processing by aggregation and filtering procedures.

Data plays an important role in DSS. They can be used directly by the user or as initial data for calculation using mathematical models.

The DSS data subsystem receives part of the data from the system for processing operations performed by the company. However, only in rare cases, the data obtained at the level of processing of commercial transactions are useful for DSS. In order to be usable, this data must be pre-processed. There are two possibilities for this. The first is to use the DBMS included in the DSS for processing data on the company's operations. The second is to do processing outside the DSS by creating a special database for this. It is clear that the second of these options is preferable for firms with a large number of commercial transactions.

ITUC. 4.5. Structure of the DSS data subsystem

The processed data about the company's operations form extractive files, which are stored outside the DSS to improve the reliability and speed of access. The idea of ​​creating a special database for the processing of the company's transactions is based on the expediency of separating the field of automatic electronic data processing from the field of a less qualified end user. In addition, the end users of the DSS, who expect the system to respond quickly to their requests, would constantly compete for machine time with the transaction processing process. Therefore, many organizations working with DSS use a separate computer operating within the central MIS to process their business transactions.

In addition to data on the operations of the firm, other internal data are required for the operation of the DSS. So, for example, estimates of managers employed in the areas of marketing, finance, production, personnel movement data, engineering data, etc. are needed. These data must be collected, entered and maintained in a timely manner.

Important, especially for decision support at the upper levels of management, are data from external sources. Required external data should include data on competitors, national and global economies. Unlike internal data, external data can often be purchased from organizations that specialize in data collection.

At present, the issue of including another source of data in the DSS is being widely studied - documents that include records, letters, contracts, orders, etc. If the content of these documents is recorded in memory (for example, on a video disk) and then processed according to some key characteristics (suppliers, consumers, dates, types of services, etc.), DSS will receive a new powerful source of information.

The data subsystem, which is part of the DSS, should have the following capabilities:

Compilation of combinations of data obtained from various sources through the use of aggregation and filtering procedures;

Quick addition or exclusion of one or another data source;

Building a logical data structure in terms of the user;

Using and manipulating informal data to experimentally test the user's working alternatives;

Data management using a wide range of management functions provided by the DBMS;

Ensuring complete logical independence of the database included in the DSS data subsystem from other operational databases operating within the company.

Model subsystem. Along with providing access to data, the DSS provides user access to decision-making models. This is achieved by introducing appropriate models into the IS and using a database in it as a mechanism for integrating models and communicating between them (see Fig. 4.6).

The resulting DSS will combine the advantages of SEOD and MIS in terms of data processing and generation of management reports with the advantages of operations research and econometrics in terms of mathematical modeling of situations and finding a solution.

The process of creating models should be flexible. It should include a special modeling language, a set of individual software blocks and modules that implement individual components of various models, as well as a set of control functions.

The use of models ensures the ability of the DSS to conduct analysis. Models, using a mathematical interpretation of the problem, with the help of certain algorithms, contribute to finding information useful for making the right decisions. For example, model linear programming makes it possible to determine the most advantageous production program production of several types of products under given constraints on resources.

The use of models as part of information systems began with the use of statistical methods and methods financial analysis, which were implemented by commands of conventional algorithmic languages. Later, special languages ​​were created that make it possible to model situations like “what if?” or “how to do it?”. Such languages, created specifically for building models, make it possible to build models of a certain type that provide a solution with a flexible change in variables.

Currently, there are many types of models and ways to classify them, for example, by purpose of use, scope of possible applications, how variables are evaluated, etc.

The purpose of creating models is either optimization or description of some object or process. Optimization models are associated with finding the minimum or maximum points of some indicators. For example, managers often want to know what their actions lead to profit maximization (cost minimization). Optimization models provide such information. Descriptive models describe the behavior of some system and are not intended for management (optimization) purposes.

Although most systems are stochastic in nature (i.e., their state cannot be predicted with absolute certainty), most mathematical models are built to be deterministic. Deterministic models evaluate variables with a single number (as opposed to stochastic models that evaluate variables with multiple parameters). Deterministic models are more popular than stochastic models because they are less expensive and difficult, and easier to build and use. In addition, often with their help it is possible to obtain sufficient information to help the decision maker.

From the point of view of the scope of possible applications, the models are divided into specialized models, intended for use with only one system, and universal models, intended for use with several systems. The first of these are more expensive, they are usually used to describe unique systems and are more accurate than the second.

Model base. Models in the DSS form a model base that includes strategic, tactical and operational models, as well as a set of model blocks, modules and procedures used as elements for building models (see Fig. 4.6). Each type of model has its own unique characteristics.

Strategic models are used at the highest levels of management to establish the goals of the organization, the amount of resources needed to achieve them, as well as the policy for acquiring and using these resources. They can also be useful for choosing options for locating enterprises, predicting competitor policies, and so on. Strategic models are characterized by a significant breadth of coverage, many variables, and presentation of data in a compressed aggregated form. Often these data are based on external sources and may be subjective. The planning horizon in strategic models is usually measured in years. These models are usually deterministic, descriptive, specialized for use in one particular firm.

Tactical models are used by middle-level managers to allocate and control the use of available resources. Among the possible areas of their use should be indicated: financial planning, planning requirements for employees, planning to increase sales, building layout schemes for enterprises. These models are usually applicable only to individual parts of the firm (for example, to the production and distribution system) and may also include aggregates. The time horizon covered by tactical models lies between one month and two years. Data from external sources may also be required here, but the main focus in implementing these models should be given to the firm's internal data. Usually tactical models are implemented as deterministic, optimization and universal.

Operational models are used at lower levels of management to support operational decision making with a horizon measured in days and weeks. Possible applications of these models include the introduction of accounts receivable and credit calculations, calendar production planning, inventory management, etc. Operational models usually use intracompany data for their calculations. They tend to be deterministic, optimizing, and generic (i.e., can be used by different organizations).

In addition to strategic, tactical and operational models, the base of DSS models includes a set of model blocks, modules and procedures. This may include procedures for linear programming, statistical analysis of time series, regression analysis, etc. - from the simplest procedures to complex application packages. Model blocks, modules and procedures can be used both individually, independently to help DSS users, and in a complex, in combination, to build and maintain models.

Interface management system. The effectiveness and flexibility of DSS in solving certain problems largely depends on the characteristics of the interface used. The interface includes software system dialogue control (CUD), the computer and the user himself.

User language - these are the actions that the user performs in relation to the system by using the capabilities of the keyboard, electronic pencils writing on the screen, joystick, mouse, voice commands, etc. The simplest form of the action language is the creation of forms of input and output documents. Having received the input form (document), the user fills it with the necessary data and enters it into the computer. The DSS performs the necessary analysis and issues the results in the form of an output document of the established form.

Increased significantly for Lately the popularity of the visual interface developed by the American company "Apple Mackintosh", which is based on the use of a special "mouse" device. Using this device, the user selects the objects and actions presented to him on the screen in the form of pictures, thus realizing the language of actions.

Controlling a computer with the human voice is the simplest and therefore the most desirable form of action language. It has not yet been sufficiently developed and, therefore, is not very popular in the DSS. Existing developments require serious restrictions from the user (a limited set of words and expressions; a special device that takes into account the characteristics of the user's voice; control should be in the form of discrete commands, and not in the form of ordinary smooth speech). The technology of this approach is being intensively improved, and in the near future we can expect the appearance of new advanced DSS using speech input of information.

The message language is what the user sees on the display screen (characters, graphics, color), data received on the printer, audio output, and so on. For a long time, the only implementation of a message language was a printed or displayed report (or other required message). Now it has been joined by a new possibility of presenting output data - computer graphics. It makes it possible to create color graphics in three dimensions on screen and paper. The use of computer graphics, which significantly increases the visibility and interpretability of output data, is becoming increasingly popular in DSS.

Over the past few years, a new direction has emerged that develops computer graphics - animation. Animation is particularly effective for interpreting DSS outputs associated with modeling physical systems and objects. So, for example, a DSS designed to serve customers in a bank, with the help of cartoon models, can realistically view various options for organizing service depending on the flow of visitors, the allowable queue length, the number of service points, etc.

In the coming years, we should expect the use of the human voice as the language of DSS messages. As a possible example, one can point to the use of this form in the work of DSS in the field of finance, where, in the process of generating emergency reports, the reasons for the exclusivity of a particular position are explained by voice.

User knowledge is what the user needs to know when working with the system. This includes not only the action plan that is in the user's head, but also textbooks, instructions, and reference data issued by the computer when the command is for help. Instructions and reference data issued by the system at the request of the user are usually not standard, but depend on the place in the context of the solution of the problem in which the DSS user is located. In other words, help is specialized in terms of the situation.

The so-called batch files containing programmed instructions for the system to execute standard procedures can be of great help to the DSS user. Such files are activated by pressing a single key and do not require knowledge of the command language from the user. An example is the procedures for comparing the planned and actual state of production (values ​​in the warehouse, production volumes, cash receipts, etc.) that are constantly performed within the framework of the automated workplace.

In case of a clear lack of user knowledge about a given subject area and the DSS itself, the latter can be used as simulators under the guidance of experienced users or experts in the field under study.

The improvement of the DSS interface is determined by the progress in the development of each of the three components indicated.

An important measure of the effectiveness of the interface used is the chosen form of dialogue between the user and the system. Currently, the most common forms of dialogue are: challenge-response mode, command mode, menu mode, and blank-fill mode in computer-supplied expressions. Each form, depending on the type of task, the characteristics of the user and the decision being made, may have its own advantages and disadvantages.

The DSS interface should have the following capabilities:

Manipulate various forms of dialogue, changing them in the decision process at the user's choice;

Transfer data to the system in various ways;

Receive data from various devices of the system in various formats;

Flexibly maintain (provide assistance upon request, suggest) the user's knowledge.

Operational requirements for DSS from the standpoint of the user.

The first three of the following requirements are related to the type of problem being solved by the decision maker. The rest are related to the type of assistance provided to him.

1. DSS should provide assistance in decision making and be especially effective in solving unstructured and poorly structured problems. This refers to tasks in which the use of SEOD, MIS, and operations research models usually did not give results.

2. DSS should provide assistance in decision-making by managers at all levels, as well as in the coordination of decisions that require the participation of several levels of management.

3. DSS should provide assistance in making both individual and collective decisions. This refers to decisions in which responsibility is divided among several managers or within a group of employees.

4. DSS should provide assistance at all stages of the decision-making process. As will be shown below, if at the stages of studying the problem and collecting data, the DSS provides only additional assistance (the main contribution is made by the use of MIS), then at all subsequent stages (except for the decision-making stage), the assistance provided by the DSS is predominant.

5. DSS, assisting in making various decisions, cannot depend on any of them.

6. Using DSS should be easy. This is ensured by the high adaptability of the system in relation to the type of tasks, the characteristics of the organizational environment and the user, as well as a friendly interface.

Group DSS

Everything that was said above about DSS was primarily related to the support of individual decisions. However, the manager rarely makes the decision alone. Boards of directors, scientific and technical councils, teams of designers, problem committees - this is not a complete list of examples of a collective approach to decision-making. Group DSS (GDSS) are interactive computer systems designed to provide support to groups of workers in solving poorly structured problems.

Group decision making is more complex than individual decision making because it involves the need to reconcile different individual points of view. Therefore, the main task of the SSPPR is to improve communication in the working team. Improved communication results in savings in labor time, which can be used to delve deeper into a given problem and develop more possible alternatives to solve it. Evaluation of more alternatives contributes to the choice of a more informed decision.

The importance of making group decisions, on the one hand, the chronic vices of group communication (see Chapter 2) and the limited ability to deal with them, on the other, led to the creation of a special information technology to support group decisions.

Much of this technology is implemented through Office Automation Systems (CAO)1, improving communication between employees. SPSS can be specialized (adapted to solve only one type of problem) or universal (designed to solve a wide range of issues). Many SPSS contain a built-in software mechanism that prevents the development of negative trends in group communication (the emergence of conflict situations, group thinking, etc.).

Structure of the SPSS. SPSS includes hardware and software, as well as procedures and personnel (see Fig. 4.7).

Rice. 4.7. Structure of a group decision support system

These components provide group members with communication and other support when discussing issues. While working with the system, team members have constant access to the database, the database of models and various applications. The manager of the group is responsible for selecting the procedures necessary for the operation of the group. The manager of the group and its members have the opportunity to enter into a dialogue.

Technical support. SPSS usually uses one of the following hardware configurations:

1. The only computer. In this case, all participants gather around a single computer and take turns answering questions that appear on the monitor screen until a solution is received. Using this configuration is only useful for educational purposes.

2. Network of computers or terminals. Each participant is at his own computer or terminal, having the opportunity to conduct a dialogue with the central processor of the system.

3. Decision room. At the heart of this GDSS configuration is the CAO1 application called computer conference and described in Section 4.4. The decision room includes a local computer network with a server running the system manager. It is also equipped with a common screen that allows you to show all members of the group the necessary information (individual and aggregated).

Software. The GSPPR software includes a database, a database of models and programs for special applications. It provides the possibility of individual and group work of users, as well as the maintenance of group decision-making procedures. So, in terms of group work, the GSPPR software allows

Perform numerical and graphic summation of proposals and voting results of group members;

Calculate the weights of decision alternatives, make an anonymous record of the proposals received, choose a group leader, build consensus building procedures, prevent the development of negative trends in group communication;

Transfer text and numeric data between group members, between group members and the group manager, and also between group members and the GPSS CPU.

Staff. This component of the GDSS includes all members of the group and a steward who is present at each meeting of the group and is responsible for the hardware of the system and managing the change in discussion procedures.

Please. Procedures are a necessary component of the GDSS, through which the purposefulness of the exchange of views, the objectivity of reaching consensus and the efficient use of the software and hardware of the system are ensured.

Support provided by SPSS. In order to analyze the work of the SDSS, we will single out three levels of support tools provided by these systems:

Level 1. Communication support

Level 2. Decision support

Level 3. Support for the rules of the game

Level 1. Communication support. At this level, SPSS, using the capabilities of CAO and special programs, can provide the following types of support:

Transfer of messages between group members by means of e-mail;

Formation of a common screen visible to all members of the group and accessible from each workplace;

Possibility of anonymous input of ideas (suggestions) and their anonymous evaluation (ranking);

Issuance on a common screen (or monitor of each workplace) of all output information that is the result of the discussion (initial and final list of proposals, voting results, etc.);

Formation of the agenda for discussion.

Level 2. Decision support. At this level, SPSS, using software tools for modeling and decision analysis, can provide the following types of support:

Planning and financial modeling;

Using decision trees;

Use of probabilistic models;

Using resource allocation models.

Level 3. Support for the rules of the game. At this level, SPSS uses special software to comply with the established rules for conducting group procedures (for example, setting the order of speeches and voting rules, acceptability of questions at the moment, etc.).

one) . Before the meeting, the group leader meets with the group facilitator to plan the group's work, select software, and set the agenda.

2). The work of the group begins with the fact that its leader offers the group a question or problem to solve.

3). Next, the participants enter their answers from the keyboard, which are made available to everyone. After the participants have familiarized themselves with all the proposals made, they give comments on them (positive or negative).

4) . The facilitator, using the proposal generalization program, searches the submitted proposals for common terms, topics and ideas and creates several generalized proposals from them with comments that are communicated to all participants.

five) . The leader initiates a discussion on generalized sentences (verbal or electronic). At this stage, with the help of special programs, the ranking (prioritization) of the proposals under discussion takes place.

6). For the top five or ten proposals, a new discussion begins to refine and further evaluate them.

7). The process (development of proposals, their generalization and ranking) is repeated or ends with a final vote. This stage uses a special program called “final comment”, which produces a comment on the selected generalized sentences.

CONSTRUCTION AND USE OF DSS FOR FINANCIAL PLANNING

The described example is based on real events that took place in one of the Western banks.

At the end of the next financial year, the bank, having found a significant decrease in profits, felt itself in danger. The analysis of the situation that had arisen went beyond the scope of ordinary managerial activity.

Although this bank was among the leading ones, one of the first to introduce credit cards and a computerized accounting system, the implementation of credit policy in it was still carried out manually.

It was decided to create a new computer system financial planning, which performs analysis and forecasting, as well as creates reports based on the use of data from the system for processing accounting operations already existing in the bank. At the same time, the analysis concerned the coverage of the dynamics of changes in the main indicators that assess the ratio of the bank's own assets and borrowed funds. Forecasting was supposed to be carried out for two constant horizons: 12 months and 5 years.

The financial planning system (FPS) was used in the following three areas:

At the beginning of each month, a report was issued on the activities of the bank for the previous month;

During each month - to solve special current problems and develop strategic plans;

At the end of each calendar year - to develop annual budget documents.

As it is easy to see, in contrast to the accounting calculations that already existed in the IS bank (which was a centralized EDMS), the newly created SFS is a DSS that retains such standard functions of these systems as

Access to data at any time;

Support for decisions made by issuing periodic management reports;

Using mathematical forecasting models to evaluate alternatives and strategies;

Ensuring the possibility of working in a dialogue mode (the possibility of changing goals and restrictions when conditions and circumstances in the financial markets change).

Data. Each month, the data obtained are recorded in databases containing retrospective information for the last three years on a monthly basis and for seven and a half years on a quarterly basis. In addition, the databases contain the received forecast information for the next 12 monthly periods.

Reports and analysis. Every month, the financial planning system produces a complete set of financial documents, including a balance sheet, income statement and reports on major commercial performance. The monthly data obtained are compared with the forecast results, the budget and with similar data obtained in the previous year. In addition, the system issues periodic reports on especially stressful (critical) aspects of the bank's activities, for example, a report on the ratio of rates and volumes of interest payments.

Forecasting. All listed reports can be issued by the system for each of the following 12 months. The explanatory variables for these reports can be entered directly by users or automatically generated for strategic reasons. If necessary, optimization models that are in the database of system models can be used here. The forecast is “rolling”, constantly covering the next 12 months, with a constant re-evaluation of the data at the beginning of each month.

Advantages. The introduction of SFP led to an increase in the profitability of the bank due to the following factors:

Building a mechanism for managing the most important indicators of the balance sheet, including liquidity and the ratio of equity and borrowed capital;

Creation of a base for coordinating the decision-making process at the level of strategic planning;

Creating the ability for senior management to respond quickly to changing regulations, market conditions and intra-bank

circumstances;

Reducing the cost of creating periodic management reports

Questions for self-examination

1. Describe the situation that prompted the bank's management to create the SFP.

2. What benefits did the introduction of SFP provide?

3. Describe the components of the SFP, justifying what type of IS it belongs to.

3). DSS has the ability to manage the dialogue between the user and the system, as well as manage data and models.

Federal State Budgetary Educational Institution of Higher Professional Education

"RUSSIAN ACADEMY OF THE NATIONAL ECONOMY

AND PUBLIC SERVICE

under the PRESIDENT OF THE RUSSIAN FEDERATION"

Northwestern Institute of Management

Faculty: State and municipal administration

Department: General Management and Logistics

Course work

"Decision Support Systems"

3rd year student

Full-time education

Fetiskin Ivan Yurievich

Work manager

Associate Professor, Candidate of Philological Sciences

Mysin Nikolay Vasilievich

St. Petersburg 2015

Introduction

Chapter 1. Theoretical aspects and concepts of decision support systems

1 Definition of a decision support system, its functions

2 Structure of decision support systems

3 Data stores

4 OLAP technologies

5 Intelligent data analysis

6 Classifications of decision support systems

7 Applications

8 DSS market

9 Decision Support System Evaluation (DSS)

Chapter 2 Practice of implementation of DSS on the example of territorial branches of the Bank of Russia

1 Formulation of the goals and objectives of the study, characteristics of the object under study

2 General overview and job description

2.1 Development of a DSS in managing the activities of the territorial branches of the Bank of Russia

2.2 Description of functional subsystems

2.3 Development of a DSS at the level of technical specifications that implements methodological and instrumental solutions

3 Conclusions and results of application of this DSS

Conclusion

Bibliography

Introduction

Developing market relations, decentralization of management, rapid obsolescence of information determine the high demands on the modern leader. Knowledge and skillful use of the provisions of management greatly facilitate the work of the head, help him to prioritize and systematize the work. Organizational structures serve as the basis on which all management activities are built.

Organizations create structures in order to ensure the coordination and control of the activities of their units and employees. The structures of organizations differ from each other in complexity (i.e., the degree of division of activities into various functions), formalization (i.e., the degree to which pre-established rules and procedures are used), the ratio of centralization and decentralization (i.e., the levels at which managerial solutions).

Structural relationships in organizations are the focus of many researchers and managers. In order to effectively achieve goals, it is necessary to understand the structure of work, departments and functional units. The organization of work and people largely influences the behavior of workers. Structural and behavioral relationships, in turn, help set the goals of the organization, influence the attitudes and behavior of employees. The structural approach is applied in organizations to ensure the basic elements of activities and the relationships between them. It involves the use of division of labor, control coverage, decentralization and departmentalization.

In the context of the dynamism of modern production and social structure, management must be in a state of continuous development, which today cannot be achieved without exploring the ways and possibilities of this development, without choosing alternative directions. Management research is carried out in the daily activities of managers and staff and in the work of specialized analytical groups, laboratories, departments. The need for management systems research is dictated by a fairly large range of problems that many organizations have to face. The success of these organizations depends on the correct solution of these problems.

The organizational structure of management is one of the key concepts of management, closely related to the goals, functions, management process, the work of managers and the distribution of powers between them. Within the framework of this structure, the entire management process takes place (the movement of information flows and the adoption of managerial decisions), in which managers of all levels, categories and professional specializations participate. The structure can be compared with the framework of the building of the management system, built to ensure that all processes occurring in it are carried out in a timely manner and with high quality.

Differences in the structure of the organization, in the features of their functioning, leave a very significant imprint on managerial activity, and in some cases have a decisive influence on it. In addition, the activities of the leader, its psychological characteristics depend not only on the type of organizational structure, but also on its hierarchical place in this structure, which, in fact, makes the topic of this course work the most relevant.

Scientifically substantiated formation of organizational management structures is an urgent task modern stage adaptation of economic entities to a market economy. In modern conditions, it is necessary to widely use the principles and methods of designing a management organization based on a systematic approach.

THE PURPOSE OF THIS COURSE WORK is to study the principle of hierarchy in the management structure of the organization.

To achieve this goal, the following tasks are defined in the work:

study of the essence and principles of building organizational structures, their classification and stages of historical development;

study of the essence and principles of building organizational structures;

building a strategy for organizational change.

RESEARCH METHODS: analytical, graphic.

To write this work, scientific works and developments of domestic and foreign authors devoted to the issues of process management, the creation of management decision support systems were used. The paper uses materials published in the Russian and foreign press, as well as those presented on specialized professional Internet sites.

Chapter 1. Theoretical aspects and concepts of decision support systems

1 Definition of a decision support system, its functions

It is obvious that the decisions made on the strategy and tactics of the development of the city must be carefully thought out and justified. This is especially important in socio-economic systems, since the decisions made concern living people, their material and spiritual condition. However, to date, decision-making by the mayor, city administration, committees is based on the experience and intuition of leaders. But socio-economic systems are complex and their behavior is difficult to predict due to the presence of a huge number of direct and feedback often not obvious at first glance. The human brain is unable to cope with a task of this dimension, so it is necessary to provide information and analytical support for decision making. In recent years, a new direction in the field of automation of managerial work has been formed and is actively used - decision support systems. They are successfully used in a variety of industries: telecommunications, finance, trade, industry, medicine and many others.

The concept of decision support systems (DSS) includes a number of tools combined common goal- to promote the adoption of rational and effective management decisions.

A decision support system (DSS) is a computer automated system, the purpose of which is to help people who make decisions in difficult conditions for a complete and objective analysis of subject activity. It is an interactive system using decision rules and corresponding models with databases, as well as an interactive computer simulation process.

DSS emerged as a result of the merger of management information systems and database management systems. DSS are human-machine systems that allow decision makers to use data, knowledge, objective and subjective models to analyze and solve unstructured and poorly formalized problems.

The decision-making process is the receipt and selection of the most optimal alternative, taking into account the miscalculation of all consequences. When choosing alternatives, one must choose the one that most fully meets the goal, but at the same time one has to take into account a large number of conflicting requirements and, therefore, evaluate the chosen solution according to many criteria.

The decision support system is designed to support multicriteria decisions in a complex information environment. At the same time, multi-criteria is understood as the fact that the results of decisions made are evaluated not by one, but by the totality of many indicators (criteria) considered simultaneously. Information complexity is determined by the need to take into account a large amount of data, the processing of which is practically impossible without the help of modern computer technology. Under these conditions, the number possible solutions, as a rule, is very large, and the choice of the best of them "by eye", without a comprehensive analysis, can lead to gross errors.

DSS also makes it possible to facilitate the work of business leaders and increase its efficiency. They significantly speed up the solution of problems in business. DSS contribute to the establishment of interpersonal contact. On their basis, training and training of personnel can be carried out. These information systems allow you to increase control over the activities of the organization. The presence of a well-functioning DSS provides great advantages over competing structures. Thanks to the proposals put forward by the DSS, new approaches to solving everyday and non-standard tasks are opening up.

DSS is characterized by the following distinctive features:

· orientation towards solving poorly structured (formalized) tasks, typical mainly for high levels of management;

· the possibility of combining traditional methods of accessing and processing computer data with the capabilities of mathematical models and methods for solving problems based on them;

· focus on the non-professional end user of a computer through the use of an interactive mode of operation;

· high adaptability, providing the ability to adapt to the features of the available hardware and software, as well as user requirements.

The decision support system solves two main tasks:

.choice best solution from the set of possible (optimization);

2.ordering possible solutions by preference (ranking).

For the analysis and development of proposals in the DSS, different methods are used. It can be:

· information search,

· data mining,

· search for knowledge in databases,

· case based reasoning

· simulation Modeling,

· evolutionary computing and genetic algorithms,

· neural networks,

· situational analysis,

· cognitive modeling, etc.

Some of these methods have been developed within the framework of artificial intelligence. If the work of the DSS is based on artificial intelligence methods, then one speaks of an intellectual DSS or IDSS.

Classes of systems close to DSS are expert systems and automated control systems.

The system allows you to solve the problems of operational and strategic management based on accounting data on the company's activities.

The decision support system is a set of software tools for data analysis, modeling, forecasting and management decision-making, consisting of the corporation's own developments and purchased software products (Oracle, IBM, Cognos).

Theoretical research in the development of the first decision support systems was carried out at the Carnegie Institute of Technology in the late 50s and early 60s of the XX century. It was possible to combine theory with practice by specialists from the Massachusetts Institute of Technology in the 60s. In the middle and late 80s of the XX century, such systems as EIS, GDSS, ODSS began to appear. In 1987, Texas Instruments developed the Gate Assignment Display System for United Airlines. This has greatly reduced losses from flights and adjusted the management of various airports, ranging from O International Airport Hare in Chicago and ending with Stapleton in Denver, Colorado. In the 90s, the scope of DSS capabilities expanded due to the introduction of data warehouses and OLAP tools. The emergence of new reporting technologies has made DSS indispensable in management.

1.2 Structure of the DSS

If we talk about the structure of the DSS, then there are four main components:

· Information data warehouses. A data warehouse is a data bank of a certain structure containing information about manufacturing process companies in a historical context. The main purpose of the repository is to provide fast execution of arbitrary analytical queries. (More details about data warehouses are discussed in paragraph 1.3 of Chapter 1.)

· Multidimensional database and analysis tools OLAP (On-Line Analytical Processing) - the service is a tool for analyzing large amounts of data in real time. (detailed in paragraph 1.4 of chapter 1)

· Data mining tools. With the help of data mining tools, you can conduct deep data mining. (More details in paragraph 1.5 of Chapter 1.)

The DSS is based on a complex of interrelated models with appropriate information support for research, expert and intelligent systems that include experience in solving management problems and ensure the participation of a team of experts in the process of developing rational decisions.

Figure 1 below shows the architectural and technological scheme of information and analytical decision support:

Fig.1 Architectural and technological scheme of the DSS

Analytical DSS systems allow solving three main tasks:

.reporting,

.real-time information analysis (OLAP),

.data mining.

3 Data stores

It is clear that decision-making should be based on real data about the control object. Such information is usually stored in the operational databases of OLTP systems. But these operational data are not suitable for the purposes of analysis, since aggregated information is mainly needed for analysis and strategic decision-making. In addition, for the purposes of analysis, it is necessary to be able to quickly manipulate information, present it in various aspects, make various ad hoc queries to it, which is difficult to implement on operational data for reasons of performance and technological complexity.

The solution to this problem is to create a separate data warehouse (DW) containing aggregated information in a convenient way. The purpose of building a data warehouse is to integrate, update and harmonize operational data from heterogeneous sources in order to form a single consistent view of the control object as a whole. At the same time, the concept of data warehouses is based on the recognition of the need to separate data sets used for transactional processing and data sets used in decision support systems. Such separation is possible by integrating detailed data separated in various data processing systems (DPS) and external sources into a single storage, their coordination and, possibly, aggregation.

The main advantages of DSS data warehouses should be noted:

· A single source of information: the company receives a verified single information environment on which all reference and analytical applications will be built in the subject area for which the repository is built. This environment will have a single interface, unified storage structures, common directories and other corporate standards, which will facilitate the creation and support of analytical systems.

· Also, when designing an information data warehouse, special attention is paid to the reliability of the information that enters the repository.

· Performance: The physical structures of the data warehouse are specially optimized to perform completely random selections, which allows you to build really fast query systems.

· Speed ​​of development: the specific logical organization of the repository and the existing specialized software allow you to create analytical systems with minimal programming costs.

· Integration: integration of data from different sources is already done, so it is not necessary to perform a data connection every time for queries requiring information from several sources. Integration refers not only to the joint physical storage of data, but also to their substantive, coordinated association; cleaning and alignment during their formation; compliance with technological features, etc.

· Historicity and stability: OLTP systems operate with up-to-date data, the period of application and storage of which usually does not exceed the value of the current business period (six months to a year), while the information data warehouse is aimed at long-term storage of information for 10-15 years. Stability means that the actual information in the data warehouse is not updated or deleted, but only adapted in a special way to changes in business attributes. Thus, it becomes possible to carry out a historical analysis of information.

· Independence: the allocation of information storage significantly reduces the load on OLTP systems from analytical applications, thus the performance of existing systems does not deteriorate, but in practice there is a decrease in response time and improved system availability.

Thus, the data warehouse operates according to the following scenario. According to a given regulation, it collects data from various sources - databases of online processing systems. The storage supports chronology: along with the current data, historical data is stored with an indication of the time to which they refer. As a result, the necessary available data about the control object is collected in one place, brought to a single format, agreed and, in some cases, aggregated to the minimum required level of generalization.

And on the basis of the data warehouse, it is already possible to draw up reports for management, analyze data using OLAP technologies and data mining (Data Mining).

The DSS reporting service helps the organization cope with the creation of all kinds of information reports, certificates, documents, summary statements, etc., especially when the number of reports issued is large and the forms of reports often change. DSS tools, by automating the release of reports, make it possible to convert their storage into electronic form and distribute them over the corporate network among company employees.

Along with large corporate data warehouses, Data Marts are also widely used. A data mart is a small specialized storage for a certain narrow subject area, focused on storing data related to one business topic. A data mart project requires less investment and is completed in a very short time. There can be several such data marts, say a revenue data mart for a company's accounting department and a customer data mart for a company's marketing department.

1.4 OLAP technologies

Interacting with the OLAP system, the user will be able to perform flexible viewing of information, obtain arbitrary data slices, and perform analytical operations of detailing, convolution, end-to-end distribution, comparison over time. All work with the OLAP system takes place in terms of the subject area. The concept of online analytical processing (OLAP) is based on a multidimensional representation of data.

The term OLAP was introduced by E. F. Codd in 1993. In his article, he considered the shortcomings of the relational model, primarily the inability to "combine, view and analyze data in terms of multiple dimensions, that is, in the most understandable way for corporate analysts", and defined General requirements to OLAP systems that extend the functionality of relational DBMS and include multidimensional analysis as one of their characteristics.

According to Codd, a multi-dimensional conceptual view is the most natural view of management personnel on the object of management. It is a multiple perspective, consisting of several independent dimensions along which certain sets of data can be analyzed. Simultaneous analysis on multiple dimensions of data is defined as multivariate analysis. Each dimension includes directions of data consolidation, consisting of a series of successive levels of generalization, where each higher level corresponds to a greater degree of data aggregation for the corresponding dimension. Thus, the dimension Contractor can be determined by the direction of consolidation, consisting of the levels of generalization "enterprise - subdivision - department - employee". The "Time" dimension can even include two directions of consolidation - "year - quarter - month - day" and "week - day", since the time count by month and week is incompatible. In this case, it becomes possible to arbitrarily select the desired level of information detail for each of the measurements. The operation of descent (drilling down) corresponds to the movement from the higher levels of consolidation to the lower ones; on the contrary, the operation of lifting (rolling up) means moving from lower levels to higher ones.

1.5 Data mining

The greatest interest in DSS is data mining, as it allows for the most complete and in-depth analysis of the problem, makes it possible to detect hidden relationships, and make the most reasonable decision. The current level of development of hardware and software tools for some time now has made it possible to maintain databases of operational information at various levels of government. In the course of their activities, industrial enterprises, corporations, departmental structures, bodies state power and local governments have accumulated large amounts of data. They contain great potential for extracting useful analytical information, on the basis of which you can identify hidden trends, build a development strategy, and find new solutions.

Data mining, IAD (Data Mining) is a decision support process based on the search for hidden patterns (information patterns) in data. At the same time, the accumulated information is automatically generalized to information that can be characterized as knowledge.

In general, the IAD process consists of three stages:

.identifying patterns;

.using the revealed patterns to predict unknown values ​​(predictive modeling);

.exception analysis, designed to identify and interpret anomalies in the patterns found.

New computer technologies that form IAD are expert and intelligent systems, artificial intelligence methods, knowledge bases, databases, computer modeling, neural networks, fuzzy systems. Modern technologies IAD allows you to create new knowledge, revealing hidden patterns, predicting the future state of systems. The main method of modeling the socio-economic development of the city is the simulation method, which allows you to explore the city system using an experimental approach. This makes it possible to play various development strategies on the model, compare alternatives, take into account the influence of many factors, including those with elements of uncertainty.

The model constructed in this work belongs to this class of systems. On its basis, local governments of strategic and tactical levels get the opportunity to analyze the dynamics of the development of a complex socio-economic urban system, identify relationships that are not obvious at first glance, compare various alternatives, analyze anomalies and make the most reasonable decision.

It is promising to use combined decision-making methods in DSS in combination with artificial intelligence methods and computer modeling, various simulation and optimization procedures, decision-making in combination with expert procedures.

1.6 DSS classifications

There are three types of DSS based on interaction with the user:

· passive ones help in the decision-making process, but cannot put forward a specific proposal;

· active participants are directly involved in the development of the right solution;

· cooperative ones involve the interaction of the DSS with the user. The proposal put forward by the system can be finalized, improved, and then sent back to the system for verification. After that, the proposal is again presented to the user, and so on until he approves the decision.

According to the method of support, there are:

· model-based DSS, use access to statistical, financial or other models in their work;

· DSS based on communications support the work of two or more users involved in a common task;

· Data-driven DSS have access to the organization's time series. They use in their work not only internal, but also external data;

· Document-oriented DSS manipulate unstructured information contained in various electronic formats;

· Knowledge-based DSS provide specialized, fact-based solutions to problems.

By area of ​​​​use, there are:

· System-wide - work with large storage systems and are used by many users.

According to the architecture and principle of operation, there are:

· Functional DSS.

They are the simplest in terms of architecture. They are common in organizations that do not set global goals and have a low level of information technology development. A distinctive feature of functional DSS is that the data contained in the files of operating systems are analyzed. The advantages of such DSS are compactness due to the use of a single platform and efficiency due to the absence of the need to reload data into a specialized system. Among the shortcomings, the following can be noted: a narrowing of the range of issues solved using the system, a decrease in data quality due to the lack of a stage for their cleaning, an increase in the load on operating system with the potential for termination.

· DSS using independent data marts.

They are used in large organizations with several departments, including information technology departments. Each specific data mart is created to solve specific problems and is focused on a specific range of users. This greatly improves system performance. The implementation of such structures is quite simple. Of the negative points, it can be noted that data is repeatedly entered into different storefronts, so it can be duplicated. This increases the cost of information storage and complicates the unification procedure. Populating data marts is quite difficult due to the fact that you have to use multiple sources. There is no single picture of the organization's business, due to the fact that there is no final consolidation of data.

· DSS based on a two-level data warehouse.

Used in large companies, whose data is consolidated into single system. Definitions and methods of information processing in this case are unified. To ensure the normal operation of such a DSS, it is required to allocate a specialized team that will serve it. Such a DSS architecture is devoid of the shortcomings of the previous one, but it does not have the ability to structure data for individual user groups, as well as restrict access to information. You may experience system performance issues.

· DSS based on a three-level data warehouse.

Such DSSs use a data warehouse, from which data marts are formed that are used by groups of users who solve similar problems. Thus, access is provided both to specific structured data and to a single consolidated information. Populating data marts is simplified by using validated and cleansed data from a single source.

There is a corporate data model. Such DSS are distinguished by guaranteed performance. But there is data redundancy, which leads to an increase in storage requirements. In addition, it is necessary to coordinate such an architecture with a variety of areas with potentially different requirements.

Depending on the functional content of the system interface, there are two main types of DSS: EIS and DSS. (Execution Information System) - information systems for enterprise management. These systems are aimed at unprepared users, have a simplified interface, a basic set of offered features, and fixed forms of information presentation. EIS-systems draw a general visual picture of the current state of the company's business performance indicators and their development trends, with the possibility of deepening the information in question to the level of large company facilities. EIS-systems - the real return that the company's management sees from the introduction of DSS technologies. (Desicion Support System) 7 - full-featured systems for analyzing and researching data, designed for trained users who have knowledge both in terms of the subject area of ​​research and in terms of computer literacy. Usually, to implement DSS systems (if data is available), it is enough to install and configure specialized software from solution providers for OLAP systems and Data Mining.

Such a division of systems into two types does not mean that the construction of a DSS always involves the implementation of only one of these types. EIS and DSS can operate in parallel, sharing common data and / or services, providing their functionality to both senior management and analytical departments of companies.

1.7 Applications

Telecommunications

Telecommunication companies use DSS to prepare and make a set of decisions aimed at retaining their customers and minimizing their outflow to other companies. DSS allow companies to carry out their marketing programs more effectively, conduct more attractive billing of their services.

Analysis of records with call characteristics allows you to identify categories of customers with similar behavior patterns in order to differentiate your approach to attracting customers of a particular category.

There are categories of clients who constantly change providers in response to certain advertising campaigns. DSS make it possible to identify the most characteristics"stable" clients, ie. customers who remain loyal to one company for a long time, making it possible to focus their marketing policy on retaining this particular category of customers.

Banking

DSS are used to better monitor various aspects of banking, such as servicing credit cards, loans, investments, and so on, which can significantly improve work efficiency.

Identification of cases of fraud, risk assessment of lending, forecasting changes in the clientele - the scope of DSS and methods of data mining. The classification of clients, the selection of groups of clients with similar needs allows for a targeted marketing policy, providing more attractive sets of services to a particular category of clients.

Insurance

The set of DSS applications in the insurance business can be called classic - it is the identification of potential cases of fraud, risk analysis, and customer classification.

The detection of certain stereotypes in insurance claims, in the case of large amounts, can reduce the number of fraud cases in the future.

Analyzing the characteristic features of cases of payments under insurance obligations, Insurance companies can reduce their losses. The data obtained will lead, for example, to a revision of the discount system for customers that fall under the identified characteristics.

The classification of clients makes it possible to identify the most profitable categories of clients in order to more accurately target the existing set of services and introduce new services.

Retail

Trading companies use DSS technologies to solve such problems as procurement and storage planning, analysis joint purchases, search for patterns of behavior in time.

Analysis of data on the number of purchases and the availability of goods in stock over a certain period of time allows you to plan the purchase of goods, for example, in response to seasonal fluctuations in demand for goods.

Often, when buying a product, the buyer acquires another product along with it. The identification of groups of such goods allows, for example, placing them on adjacent shelves in order to increase the likelihood of their joint purchase.

The search for patterns of behavior in time gives an answer to the question "If today the buyer has purchased one product, then after what time will he buy another product?". For example, when purchasing a camera, a customer is likely to purchase film, develop and print services in the near future.

The medicine

There are many expert systems for making medical diagnoses. They are built mainly on the basis of rules describing combinations of various symptoms of various diseases. With the help of such rules, they learn not only what the patient is sick with, but also how to treat him. The rules help to choose the means of medication, determine the indications - contraindications, navigate the treatment procedures, create conditions for the most effective treatment, predict the outcomes of the prescribed course of treatment, etc. Data Mining technologies make it possible to detect patterns in medical data that form the basis of these rules.

Molecular genetics and genetic engineering

Perhaps the most acute and at the same time clear task of discovering regularities in experimental data is in molecular genetics and genetic engineering. Here it is formulated as a definition of the so-called markers, which are understood as genetic codes that control certain phenotypic features of a living organism. Such codes may contain hundreds, thousands, or more related items.

Large funds are allocated for the development of genetic research. Recently, there has been a particular interest in the application of Data Mining methods in this area. Several large firms are known to specialize in the application of these methods for deciphering the human and plant genomes.

Applied chemistry

Data mining methods are widely used in applied chemistry (organic and inorganic). Here, the question often arises of elucidating the features of the chemical structure of certain compounds that determine their properties. This task is especially relevant in the analysis of complex chemical compounds, the description of which includes hundreds and thousands of structural elements and their bonds.

1.8 DSS market

In the DSS market, companies offer the following types of services for the creation of decision support systems:

· Implementation of pilot projects on DSS systems in order to demonstrate to the Customer's management the high-quality potential of analytical applications.

· Creation together with the Customer of fully functional DSS systems, including data warehouse and Business Intelligence tools.

· Data warehouse architecture design, including storage structures and management processes.

· Creation of "data marts" for the selected subject area.

· Installing and configuring OLAP and Business Intelligence tools; their adaptation to the requirements of the Customer.

· Analysis of statistical analysis tools and "data mining" to select software products for the architecture and needs of the Customer.

· Integration of DSS systems into the Customer's corporate intranets, automation of electronic exchange of analytical documents between storage users.

· Development of Executive Information Systems (EIS) for the required functionality.

· Services for the integration of databases into a single information storage environment

· Training of the Customer's specialists in data warehousing and analytical systems technologies, as well as in working with the necessary software products.

· Provision of consulting services to the Customer at all stages of design and operation of data warehouses and analytical systems.

· Complex projects for the creation/modernization of the computing infrastructure that ensures the functioning of the DSS: solutions of any scale, from local systems to systems of the scale of an enterprise/concern/industry.

1.9 Decision Support System Evaluation (DMSS)

Criteria for evaluating DSS. The system must effectively manage income and risk under any market conditions, generating effective market entry and exit signals. At the same time, the frequency of transactions should be moderate, taking into account transaction costs, commissions, losses on the spread, etc. The complexity of the construction should not be intimidating. Most of those who reject numerical methods in favor of their "intuition" end up with below average results.

Naturally important characteristic in the evaluation of the system is the total (final) profit. With high operating costs, such a characteristic as profit per operation becomes important. The accuracy of decisions (percentage), calculated as the ratio of the number of profitable operations to the total number of operations, is a popular characteristic for many traders, although its importance is overestimated. The fact is that many efficient systems make wrong decisions more often than correct ones, while many non-profitable (or almost non-profitable) systems make correct decisions more often.

The maximum loss of own funds is an important characteristic for measuring the risk of the strategies used by the system. Systems subject to periodic large losses cannot be considered usable, even if, in the end, they give a sufficient net profit. At the same time, maximum losses mean not just the largest amount of losses from a sequence of unprofitable operations, but the maximum decrease in capital during the period under review. During such a decline, the sequence of losing trades may be interrupted by individual profitable trades that are not able to change the overall unprofitable nature of the period of inefficiency of the system. The main characteristic of the efficiency of the system is calculated as the ratio of the total profit to the amount of capital loss during the period of maximum inefficiency of the system and is usually called the return/risk ratio. There are also many other evaluations of the effectiveness of the system, sometimes quite complex, requiring a large amount of statistical calculations, but in most cases the above simple characteristics are sufficient. It should be noted that when evaluating the system, you can use the criteria recommended by classical theory portfolio management.

System optimization consists in finding the best formula for the indicator - the best one in terms of getting the maximum and/or most stable profit with it from data collected over a long period of time. This optimization is self-contradictory. Its critics will immediately point out that future prices may behave differently than they have in the past. Proponents of such optimization must be convinced of the existence of certain patterns, stability in price behavior that does not change or changes slightly over time.

To test the effectiveness of the fact that the rules used in technical analysis give a stable profit in the future, being themselves calculated from past data, the following simple testing method is used (the so-called blind modeling). First, the decision rule is optimized on past data, and then it is tested on later (recent) data. In this way, you can determine how well you can generally predict the future from past data using a given rule. If an indicator with optimal parameters performs well on more recent data, one can hope that it will perform well in the future.

When re-evaluating system parameters, one should proceed to new system only if the resulting "improvement" is statistically significant.

Robert Pelletier recommends limiting the number of parameters when constructing decision rules, since their increase increases the number of degrees of freedom of the system. In addition, there may be connections between them, i.e. they may turn out to be statistically dependent, which is usually seen from their cross-correlation coefficient. Pelletier believes that a good system should contain no more than 2-5 parameters.

The sample for checking the indicator should be large enough so that there are at least 30 signals for the selected period. In this case, the period should include an integer number of complete long (low-frequency) cycles in order to limit the impact of biases in the direction of sales or purchases. So, for example, for a known 4-year cycle of the stock market, the analysis should be carried out on data of at least 8 years.

organizational bank intellectual data

Chapter 2

1 Formulation of the goals and objectives of the study, characteristics of the object under study

Currently Central bank Russian Federation(hereinafter referred to as the Bank of Russia) is the key regulator of the Russian banking system and in many respects is the guarantor of its stability and economic stability. The system of the Bank of Russia has a complex organizational structure- the central office (hereinafter referred to as TA), territorial offices (hereinafter referred to as TU), and has more than 80 thousand employees. In turn, territorial institutions have in their subordination a network of cash settlement centers and other units that ensure the activities of the TC. The presence of a complex organizational structure determines the complexity of the management system of the Bank of Russia, which covers two levels - TC and CA. At present, the following main tasks are relevant for the Bank of Russia: a general reduction in costs, standardization of the activities of territorial institutions, and improvement of the management system of territorial institutions.

The process approach to management is considered as the main tool for fulfilling these tasks, the implementation of which was started in the Bank of Russia in 2002. Process approach is the prevailing approach to building a flexible and efficient management system, which has become widespread in the world over the past 10-15 years. The process approach presupposes a clear formulation of the goals and strategy of the activity, a description of the activity in the form of a set of interrelated processes that have specific results at the output, a clear distribution of responsibility between all participants in the processes.

As shows world practice, the effective application of the process approach is largely determined by the presence of an information-computing system that generates and provides the information necessary for decision-making. With the help of such a system, at the level of the technical specifications of the Bank of Russia, it would be possible to describe and control the execution of processes, evaluate their cost, calculate the real load, conduct a reasonable assessment of the effectiveness of processes, employees, departments, etc. At the level of the CA of the Bank of Russia, the system would make it possible to compare technical specifications for various indicators accumulated in the course of work, standardize technical specifications, describe process standards, replicate them in technical specifications, and solve a number of other tasks.

All of the above determines the relevance of the topic of this chapter, devoted to the development of methodological, mathematical and software-instrumental approaches to creating a decision support system in the field of managing the activities of the territorial institutions of the Bank of Russia based on the process approach (hereinafter referred to as the System, DSS "Process Management").

The purpose of this work is to develop a comprehensive methodological, mathematical, informational, software and instrumental support for a decision support system in the tasks of managing the activities of the territorial institutions of the Bank of Russia, including the level of technical specifications and the central office.

2 General overview and job description

2.1 Development new concept DSS in managing the activities of territorial branches of the Bank of Russia

The specificity of the Bank of Russia was analyzed, which consists in the presence of a complex organizational structure, a vertical two-level system for managing territorial institutions, a clear regulation of activities based on a large-scale regulatory framework, the complexity of document management, financial management features, informatization and security requirements. information security. As a result, it was found that the existing products are not fully suitable for solving the problems of managing the territorial branches of the Bank of Russia.

The study of the specifics of the Bank of Russia and the analysis of the main tasks in the management of the activity of technical institutions made it possible to formulate the following conceptual principles for constructing a DSS:

) Two-level structure. The developed DSS should function at two levels - TU (regional) and CA (federal). At the regional level, the DSS supports the management of the activities of technical specifications based on a process approach; at the federal level, information is collected on activities from all technical specifications, centralized storage and analysis of this information, classification of technical specifications, and the formation of standards;

) Full cycle management based on the process approach. For effective and continuous improvement of activities, an important characteristic of DSS is to provide a full cycle of management based on a process approach, which involves iterative execution of procedures for describing processes, monitoring and controlling execution, analyzing processes, and reengineering.

Taking into account the two-level structure of the system, the control cycle is presented in the following form (Fig. 2):

Rice. 2. Management support cycle in DSS

)Integration of approaches and technologies. In order to most effectively solve the problems of improving the activities of technical specifications in the created DSS, it is necessary to integrate the approaches and technologies of business process management (BPMS), performance management (CPM) and business intelligence (BI). These approaches should be implemented on the basis of unified architectural principles and operate within the framework of a unified information, software and technological infrastructure;

)Support of standards is necessary to solve the problems of standardization of the activities of TU. At the federal level - development, debugging, analysis of process standards, etc.; at the regional level - "imposition" of standards on existing processes;

)Integration of processes in the data warehouse. BPMS class systems are transactional and do not require a data warehouse. In the Bank of Russia, it is required not only to organize process management, but also to ensure their comprehensive analysis - dynamic, comparative, structural, etc. Therefore, information about activities should be accumulated in data warehouses of each technical institution, part of the data will be transferred to the federal level (to a centralized repository);

)Development of the methodological base for analysis. For a more complete and effective solution of the problems of analyzing information on the activities of technical specifications, it is necessary to develop a methodological and instrumental base in the following areas: calculation of the cost of processes, assessment of the duration of processes, analysis of the organizational structure, performance management;

)Interaction with TPK. DSS should interact with standard software systems (TPC) operating in territorial institutions. The interaction is organized with the aim of: obtaining initial data (for example, data on the costs of technical specifications); obtaining up-to-date regulatory and reference information; obtaining data on the execution of processes. Taking into account these principles, a conceptual model of the system was developed, covering the federal and regional levels of government (Fig. 3):

Rice. 3. Conceptual model of the DSS in managing the activities of the territorial branches of the Bank of Russia

The presented conceptual model most fully corresponds to the solution of the management tasks of the Bank of Russia and includes the following components:

· Regional level systems (in each territorial institution). The DSS at the regional level is replicable and provides functionality that is common to all technical specifications. Information about the activities of TS is accumulated in a data warehouse, over which analytical BI tools operate.

· Federal level system (in the central office). The federal-level DSS is an integrating component that involves centralized storage and processing of information about the activities of all technical specifications and functionality that is different from the regional-level system. The federal level system generates data (process standards, regulations, etc.) that are replicated in the regional level DSS.

· External sources of information mainly provide DSS data at the regional level, they include various software systems operating in territorial institutions. External sources can be considered as external components of the DSS.

Since the federal level system is largely based on data transmitted from regional level systems, it is first necessary to develop information, mathematical and instrumental support for the regional level system as the basis for an integral DSS of the Bank of Russia. At the same time, it should be noted that the developed methods and tools will be used in the construction of the federal level system. In the course of the study, the structure of the DSS at the regional level was developed (Fig. 4), while taking into account the scale of technical specifications, the variety of functions and processes performed, the factors of established management practice, and the features of current automation.

Rice. 4. Structure of DSS at the regional level of the Bank of Russia

2.2.2 Description of functional subsystems

The system includes functional subsystems that provide user interfaces and implement business functions, and technological subsystems that ensure the operation of functional subsystems based on unified data management mechanisms and centralized metadata. All subsystems operate under the control of the administration and information security subsystem, which ensures the proper level of data protection from unauthorized access in accordance with the requirements of the Bank of Russia. In the course of the study, taking into account the specifics of the Bank of Russia, the requirements for information and instrumental support of functional subsystems were developed and justified.

The process description subsystem is intended for a formalized description of activities in the form of a set of interrelated processes, taking into account the features of the Bank of Russia. To model the processes in the system, the IDEF0 and IDEF3 standards were used, supplemented by a number of additional structures: control operations, return transitions, links to other processes, auxiliary processes, process start and end points. When forming an information model for describing TS processes, the specifics of the Bank of Russia and the requirements of standards, as well as the following principles were taken into account:

· Versioning support implies maintaining a chronology of all changes in the process description (changes to objects are recorded as versions sorted by date). Due to this, it is possible to obtain a model of the activity of technical specifications as of any date;

· Support for change modeling is provided by maintaining temporary versions of objects that can be approved or revoked as needed;

· Customizability of process models involves expanding the set of process model attributes, introducing new objects and linking them with existing ones.

Taking into account the stated principles and features, an information model of processes and objects of their environment was developed in the course of the study (Fig. 5).

Rice. 5. Interrelation of the main objects of the process environment.

Based on the generated information model, the process description subsystem allows solving the following main tasks:

· the formation of a holistic formalized model of the activity of TU;

· keeping information about activities up to date;

· generation of reports and certificates on documenting the activities of TU.

The process execution control subsystem ensures the execution of formalized processes, routing tasks between performers in accordance with the description, monitoring compliance with deadlines and performance efficiency, transforming data on the execution of processes from external sources into a single unified format.0

As a result of the study, the life cycle of processes and operations was developed (Fig. 6), which, together with the notation of the description of processes, provides the solution of the following tasks:

· organization of process executions;

· monitoring and managing the execution of processes;

· organization of control over the execution of processes at critical points;

· formation of analytical reports for managers of various levels of technical specifications (heads of sectors, departments, departments, top management).

Rice. 6. Life cycle process execution

The process cost subsystem is designed to calculate the cost characteristics of processes and analyze them in various sections, provides tools for a detailed analysis of the cost characteristics of processes, balancing, comparative analysis, various options calculation.

The activity analysis subsystem implements support for the analysis of the activities of the TU in various aspects - efficiency, costs, personnel, processes, etc., while collecting and structuring data from external sources and other subsystems. The analytical subsystem is built on the basis of the CPM methodology, taking into account the tasks of the Bank of Russia and provides a set of analytical applications and tools for solving the following tasks:

.Management of the system of strategic goals, objectives and indicators (taking into account the targets set at the federal level by the Bank of Russia);

.Support for decision-making in the field of personnel management and organizational structure of TU;

.Monitoring and analysis of performance indicators.

The system of strategic goals, objectives and indicators is a system of balanced scorecards (BSC) and key performance indicators that can be set for processes, departments, employees, etc. All goals, objectives and indicators are chronological in nature. The data source for the BSC is the data warehouse. Target values ​​of indicators can be set by several scenarios, to assess the degree of achievement of goals and objectives, indicators can be assigned weighting factors. Based on a comparison of target and actual values, monitoring and analysis of the achievement of goals is carried out.

Decision support in personnel management includes analytical applications for the analysis of the organizational structure, personnel analysis in terms of performance discipline, performance and key performance indicators of processes, balancing and distribution of functional responsibilities.

Monitoring and analysis of performance indicators is carried out using BI-tools based on the repository, while providing the ability to compare heterogeneous indicators and various types of analysis (dynamic, structural, comparative, cluster, ranking, etc.).

2.2.3 Development of a DSS at the level of technical specifications that implements methodological and instrumental solutions

During the development of the DSS, the analysis of the requirements for building the system was carried out, the logical and physical data structure was developed, the basic principles of building the system were substantiated, and the tasks of choosing information technologies for the implementation of the system were solved.

The structure of the system includes functional subsystems that implement business logic and user interface, and technological subsystems that ensure the operation of functional subsystems based on unified data management mechanisms and centralized metadata.

The following information technologies have been chosen to implement the system:

· as a basis for storing information - the Oracle relational database management system version 9i;

· as a software and tool development environment - the analytical complex "Prognoz-5", focused on the development of information and analytical systems and decision support systems in various areas of the economy;

· for developing web components - the Microsoft Visual Studio 2005 integrated environment and the ASP.NET platform.

During the creation of the DSS, a set of software and technological solutions is developed based on unified architectural principles for the most optimal and reliable operation. When developing procedures for managing a complex database, including transactional and analytical segments, the following solutions were developed and applied:

· Ensuring the data consistency of the transactional and analytical segments of the database, for this a system of interconnected classes has been developed, focused on the use of a unified transaction processing core, which is based on the use of Oracle DBMS metadata. At the table level, data integrity control is provided by DBMS tools to improve the reliability of operation (Fig. 7):

Rice. 7. Scheme for managing the consistency of DSS data.

· Support for object versioning while maintaining integrity control at the DBMS level. To do this, each object is stored in two tables: a table of objects and a table of object versions;

· Database scalability at the level of attributes and objects with integrity control. For additional attributes, integrity is controlled at the trigger level; when new objects are created in tables, unified integrity control triggers are automatically created;

· Optimization of extraction and writing to the database with large amounts of data. After the creation of the physical structure, it was indexed, for the tables of the data warehouse, the tools for forming Partitions of the Oracle DBMS were used.

The sources of data for the initial filling of the DSS and subsequent updating can be data from the standard software systems operated in the TU: On-farm Activity Systems (IEA), Document Management Systems, Automation Systems, etc. The DSS allows you to download process descriptions from MS Word and Excel files, which important for territorial institutions that have draft process models "on paper".

The developed DSS is used in industrial mode in the National Bank of the Republic of Bashkortostan at more than 300 workplaces of managers and specialists to describe processes, organize and monitor the execution of processes, justify changes in the organizational structure, and analyze activities. About 980 processes are described in the system, about 730 of them are approved, about 200 processes are regularly launched in industrial mode.

2.3Conclusions and results of application of this DSS

The following main results and conclusions have been obtained:

On the basis of the findings, the concept of an integrated decision support system in the management of TS activities is presented, focused on the integration of BPMS, BI and CPM approaches, in which the methods and algorithms developed by the author are built on the basis of a single information and instrumental environment. The concept combines both new and previously known methods for monitoring and analyzing the activities of technical institutions based on a process approach, adapted to the specifics of the Bank of Russia.

A decision support system was created and tested in specific technical specifications of the Bank of Russia in the field of managing the activities of a territorial institution at the regional level. The use of DSS in technical specifications makes it possible to increase the manageability of activities based on a process approach, improve the internal control system, optimize the existing organizational structure, and form a repository based on performance indicators.

As a result of the implementation of the system, the following results were achieved (as follows from the reports to the management of the Bank of Russia):

· improved system of internal control of activities;

· technologies for issuing and cash transactions have been improved and labor costs have been reduced (up to 10% for some transactions);

· the centralization of the functions performed by the Settlement and Cash Centers (13 functions in 9 processes);

· the cash circulation department was transformed into two independent departments;

· redistributed positions between departments within the security and information protection department;

· reduction of staff in the economic and operational department was carried out; proposals are being prepared to optimize the workflow.

Conclusion

To date, there is no recognized leader in the production of software for building DSS systems. None of the companies produces a ready-made solution, which is called "out of the box", suitable for direct use in the customer's production process. The creation of a DSS always includes the stages of analyzing the customer's data and business processes, designing storage structures taking into account his needs and technological processes.

Considering the amount of financial and other resources involved, the complexity and multi-stage nature of projects for building DSS systems, the high cost of design errors is obvious. Mistakes in software selection can result in financial costs, not to mention increased project time. Data structure design errors can lead to both unacceptable performance and the cost of time spent reloading data, which sometimes reaches several days. Therefore, having a deep understanding of the architecture of data warehouses, it is necessary to avoid any mistakes, which entails a significant reduction in project execution time and the ability to get the most out of the implementation of DSS.

It should be noted separately that the problems of decision-making, namely, DSS are poorly developed in our country and are little used in practice. The use of programs such as the one described here is not only very simple, but also quite effective and does not require special knowledge and investment.

Several dozens of different firms produce products capable of solving certain problems that arise in the process of designing and operating DSS systems. This includes DBMS, tools for unloading / transforming / loading data, tools for OLAP analysis and much more.

Self-analysis of the market, the study of at least a few of these tools is not an easy and time-consuming task.

So, in this work, we got acquainted with decision support systems.

In the introduction, the relevance of this topic is substantiated, the purpose and objectives of the study are given, general characteristics work, the subject of the study was identified.

The first chapter provides theoretical aspects and concepts of decision support systems, provides a detailed classification of the types of DSS, and initially disclosed their functions. Also in this chapter, we got acquainted with the history of the creation of support systems, analyzed in more detail the structure of the DSS and its main elements. Are given distinctive features decision support systems, as well as the areas and areas in which they can be applied.

A decision support methodology has been identified, and this allows us to conclude that its application makes it possible to:

· formalize the process of finding a solution based on the available data (the process of generating solution options);

· rank the criteria and give criteria-based assessments of physical parameters that affect the problem being solved (the ability to evaluate solutions);

· use formalized coordination procedures when making collective decisions;

· use formal procedures for predicting the consequences of decisions made;

· choose the option that leads to the optimal solution to the problem.

It follows from this that we have become familiar with the basic things and the theoretical part about decision support systems.

The second chapter presents the practical implementation of the DSS in the field of managing the organization's activities based on the process approach (on the example of the territorial offices of the Bank of Russia). The concept of building a DSS "Management of the activities of the territorial institutions of the Bank of Russia" is proposed. A conceptual model of the DSS, functional structure and requirements for the main components have been developed and substantiated. A set of methods and tools is proposed to support decision-making in the management of TS, taking into account the specifics of the Bank of Russia. The requirements for the information and analytical support of the system were developed and substantiated, taking into account the urgent tasks of managing the territorial branches of the Bank of Russia. The results of the introduction of this system based on reports to the management of the Bank of Russia are given.

Thus, we found out how these decision support systems are applied in practice - in our case, in the banking sector.

The use of DSS is promising if only because any management decision is subjective, based on the company's policy, reflects the main goals of the organization and, most importantly, is not necessarily true. All this leads to the need to formalize the decision-making process and attract supportive tools to reduce the risk of making the wrong decision. The latter increases with the accumulation of information to be processed. This happens because a person is either not able to process all the necessary information to make a decision on their own, or is not able to do it in a time frame when the task is still relevant.

Bibliography

1.Vesnin, V.R. Management: Textbook - 4th ed., Revised. and additional - M.: TK Velby, 2009. - 342 p.

2.Gerchikova, I.N. The process of making and implementing managerial decisions / I.N. Gerchikova // Management in Russia and abroad, 2013. No. 12. - 130 p.

.Goncharov, V. I. Management: tutorial/ V. I. Goncharov. - Minsk: Modern School, 2010. - 255 p.

.Drobyshev, A.V. Decision-making methods. Delphi and ELECTRA methods. - Guidelines to laboratory work on the course "Decision Support Systems". - MGIEM. Comp.: I.E.Safonova,., K.Yu.Mishin, S.V.Tsyganov: M., MGIEM, 2008. - 26 p.

.Evlanov, A. G. Theory and practice of decision making. - M.: Economics, 2010. - 212 p.

.Korotkov, E. M. Management: a textbook for bachelors / E. M. Korotkov. Moscow: Yurait, 2012.- 85 p.

.Krivko, O.B. Information Technology. Moscow: SOMINTEK. 2011. - 179 p.

.Lafta, J. K. Efficiency of organization management. - M.: Russian business literature, 2009. - 320 p.

.Lafta, J. K. Efficiency of organization management. - M.: Russian business literature, 2011. - 320 p.

.Makarov, S.F. Manager at work. - M.: FINPRESS, 2009. - 155 p.

.Meskon, M. Fundamentals of Management: Textbook / M. Meskon, M. Albert, F. Hedouri; M., 2012. - 387 p.

.Pankrukhina, A.P. Control theory: textbook / [Yu. P. Alekseev and others]; under the general editorship of: A. L. Gaponenko, A. P. Pankrukhina. - Moscow: Publishing house of the RAGS, 2010.- 213 p.

.Pirozhkov, V.A. On the implementation of the process approach to management in the form of a decision support system "Management of the organization's activities" [Text] / V.A. Pirozhkov // Bulletin of the Tambov University. Ser.: Humanities. - 2008. - Issue. 11. - 489 p.

.Polushkin, O.A. Strategic management: lecture notes. - M.: EKSMO, 2007. - 138 p.

regional authorities // Reforms in Russia and problems

.Romashchenko, V.N. Decision making: situations and advice. - Kyiv, 2012. - 154 p.

16.Rumyantseva Z.P. Management of the organization: a textbook. - M.: INFRA-M, 2005. - 432 p.

.Saraev, A. D., Shcherbina O. A. System analysis and modern information technologies // Proceedings of the Crimean Academy of Sciences. - Simferopol: SONAT, 2009. - 136 p.

.Safonova, I.E. Decision-making methods. Modification of the Delphi method and the method of analysis of hierarchies. - Guidelines for laboratory work on the course "Decision Support Systems". - MGIEM. Comp.:. 18. I. E. Safonova, A. V. Drobyshev, K. Yu. Mishin, S. V. Tsyganov: M., MGIEM, 2007. - 20 p.

.Safonova, I.E. Decision-making methods. Minimum distance method and methods MaxiMin and MaxiMax. - Guidelines for laboratory work on the course "Decision Support Systems". - MGIEM. Comp.:, 18. I.E. Safonova A.V. Drobyshev, K.Yu. Mishin, S.V. Tsyganov: M., 2007. - 19 p.

.Terelyansky, P.V. Decision support systems. Design experience: monograph / P.V. Terelyansky; VolgGTU. - Volgograd, 2009. -127 p.

.Chernyakhovskaya L.R. Decision support in strategic enterprise management based on knowledge engineering / L. R. Chernyakhovskaya et al. Ufa: Academy of Sciences of the Republic of Belarus, Gilem, 2010. - 128 p.

Send your good work in the knowledge base is simple. Use the form below

Students, graduate students, young scientists who use the knowledge base in their studies and work will be very grateful to you.

Similar Documents

Study technological process for the production of aerated concrete. Model "how it will be" of the process of diagnosing the state of the technological process of production of aerated concrete, taking into account the decision support system. Prototyping of the DSS interface.

thesis, added 06/17/2017


Studying the purpose and main tasks that Project Expert solves - a decision support system (DSS) designed for managers designing a financial model for a new or existing enterprise. Software applications, stages of work.

abstract, added 05/19/2010


Classification of information systems for managing the activities of an enterprise. Market analysis and characteristics of Business Intelligence class systems. Classification of decision-making methods used in DSS. Choosing a business intelligence platform, comparison criteria.

thesis, added 09/27/2016


Classification of decision support systems. Comparative analysis methods for risk assessment of retail lending. The structure of the decision support system, the formation of the initial knowledge base. Designing an information system database.

thesis, added 07/10/2017


The concept of decision support systems. The scope of Analytica 2.0. Quantitative modeling software. Graphical interface for model development. Basic modeling methods. Impact diagram and decision tree.

control work, added 09/08/2011


Development of algorithmic and software for solving the problem of decision support on the release of new products. Mathematical support of the task of supporting decision-making on the release of new products, the main input and output data.

thesis, added 03/08/2011


Types of administrative information systems: reporting systems, decision support systems, strategic decision support systems. Sorting and filtering lists in Microsoft Excel. Working with databases in Microsoft Access.

test, added 11/19/2009