Product quality management and management decision making. Modern problems of science and education

1

The paper examines the problem of decision support in product quality management. To effectively solve it, a conceptual approach to the study of consumer satisfaction is proposed based on an in-depth analysis of consumer reviews posted on the Internet in natural language using artificial intelligence (AI) technologies. The AI ​​tools used are text mining, sentiment analysis, aspect sentiment analysis, data mining and machine learning. Specialized Internet resources are used as data sources to accumulate consumer reviews, for example, tophotels.ru. Based on the developed approach, a prototype of a decision support system has been implemented, which allows for qualitative and quantitative research of customer satisfaction. To evaluate the effectiveness of the proposed approach, an experiment was conducted to conduct a quantitative and qualitative study of hotel customer satisfaction. The results obtained indicate the effectiveness of the proposed approach to decision support in product quality management and the prospect of its use instead of classical methods of quantitative and qualitative research of consumer satisfaction.

sentiment analysis

decision support

product quality management

1. GOST R 54732-2011/ISO/TS 10004:2010 Quality management. Customer satisfaction. Monitoring and Measurement Guidelines. – M.: Standartinform, 2012. - 28 p.

2. Internet resource dedicated to the resort and hotel business [Electronic resource] - Access mode: http://www.tophotels.ru.

3. Pazelskaya A., Soloviev A. Method for determining emotions in texts in Russian // Computer linguistics and intellectual technologies. Sat. scientific articles / Vol. 10 (17). - M.: Publishing house of the Russian State University for the Humanities, 2011. - P. 510-522.

4. Yusupova N. I., Bogdanova D. R., Boyko M. V. An approach to the use of sentiment analysis in texts in Russian based on machine learning // IMMM 2012: Second International Conference “Advanced Technologies for Information Extraction and Management”, Venice, Italy. - 2012. - P. 8-14.

5. Yusupova N. I., Bogdanova D. R., Boyko M. V. Algorithmic and software for sentiment analysis of text messages using machine learning // Vestnik UGATU. - 2012. - T. 16, No. 6(51). - pp. 91-99.

6. Nasukawa T., Yi J. Sentiment analysis: capturing favorability using natural language processing // In Proceedings of the 2nd international conference on Knowledge capture, Florida, USA, October 23–25, 2003. - pp. 70–77.

7. Pang B., Lee L. Opinion Mining and Sentiment Analysis. Foundations and Trends in Information Retrieval. - Vol. 2, Nos. 1-2. - 2008. -135 p.

8. Pang B., Lee L. Thumbs up? Sentiment Classification using Machine Learning Techniques // Proceedings of the Conference on Empirical Methods in Natural. Language Processing (EMNLP). -Philadelphia. - 2002. - P. 79-86.

Introduction

To ensure product quality, an enterprise needs to make effective management decisions. The development of management decisions and their adoption should be based on knowledge and patterns obtained during the analysis of the collected information. For enterprises, such information is information on the degree of consumer satisfaction (CS), which is expressed in the form of consumer opinions about the quality of products. Therefore, when managing quality, the key information on which the adoption of certain decisions depends is customer satisfaction.

To collect data and evaluate PM, the international quality standard ISO 10004 recommends using the following methods: personal interviews, telephone interviews, discussion groups, correspondent (mailing questionnaires) research and online surveys (questionnaires). A common disadvantage of the recommended methods is the need to perform a large amount of manual work: preparing survey questions, selecting a respondent base, sending out a questionnaire and collecting results, conducting personal interviews, preparing a report on the results. All this leads to higher research costs. Also, due to their characteristic discrete nature, the methods do not allow continuous monitoring of the level of UE. For this reason, data analysis is limited to one time slice and does not provide insight into the trends and dynamics of PM. The timeliness of making management decisions also suffers, which depends on the frequency of receipt of relevant data on consumer opinions.

Questions arise from existing scales for assessing PM and the subjectivity of their perception by respondents. The resulting PM assessments are expressed in the form of abstract satisfaction indices, which are difficult to understand, compare and interpret the results. Methods for analyzing collected data recommended by the ISO 10004 standard allow identifying only linear relationships.

The purpose of the study is to increase the efficiency of the process of developing and making decisions when managing the quality of goods and services. In this paper, to improve the efficiency of product quality management, we propose an approach to decision support based on PM research using AI technologies.

Proposed approach to decision making

The proposed approach to support management decision-making in product quality management is implemented by combining methods and algorithms for collecting and processing data on PM into a unified intelligent decision support system (ISDS). The scheme for managing product quality, developed with the help of ISDS, is presented in Figure 1. The main actors in the quality management process: consumers of products and the decision maker (DM). The object of management is understood as the quality of the services provided, on which the efficiency of the enterprise depends. The subject of management is understood as the decision-maker who makes decisions to improve the quality of services. Controlling influence refers to management decisions that influence the quality of the service, for example, modernization of goods, selection of counterparties, determination of pricing policy, selection of personnel of a given qualification, introduction of innovations, increasing the level of service, provision of resources, etc. The process of developing and making decisions is ensured through intellectual decision support systems (DSSS). According to the scheme, an enterprise produces goods or services and supplies them to consumers. After using them, consumers publish their opinions about the quality of the products on the Internet in the form of text reviews. IDSPR automatically collects reviews, cleans them and loads them into the data storage subsystem. The IDSPR data analysis subsystem processes reviews, evaluates software, and extracts knowledge. The results of the review analysis are visualized in the user interaction subsystem. Based on the results obtained, the decision maker makes management decisions to improve the quality of service.

Figure 1 - Service quality management scheme using ISDS

Figure 2 shows the operating algorithm of the IDSPR. It consists of four main stages. The first stage contains procedures for collecting reviews from Internet resources, cleaning data and loading into the database. At the second stage, the collected reviews are processed and analyzed. It includes marking reviews by their emotional tone (for example, negative and positive), identifying aspects of a product, and determining the tone of individual statements about aspects. After the data processing stage using visualization tools, a quantitative study of the software program is carried out. Qualitative research on PM is carried out by constructing models based on decision trees, where the sentiment of the review acts as the dependent variable, and sentiment statements on product aspects are the independent variables. Based on the conducted research, management decisions are developed and made.

Figure 2 - Algorithm of operation of an intelligent decision support system

Applied artificial intelligence technologies

Data collection. Today there are a large number of Internet resources where users can leave their reviews about products and services. The most popular examples are tophotels.ru (635 thousand reviews), yelp.com (53 million reviews), tripadvisor.com (travel, 130 million reviews). Their advantage as a source of data for evaluating software programs lies in their purpose—the accumulation of consumer reviews. Unlike social networks, Internet database pages use XML markup, which defines the structure specific to the review. This structure contains separate blocks with the name of the product or enterprise, with a review and other blocks with additional information. This greatly simplifies the data collection process and eliminates the problem of keyword ambiguity. Another advantage is that on many such resources there is moderation of reviews and confirmation of the author’s objectivity.

There are two main ways to collect data from online consumer review resources: 1) using an API (application programming interface) and 2) web parsing. An API is a set of ready-made tools - classes, procedures, functions provided by an application (Internet resource) for use in external software products. Unfortunately, few resources that accumulate reviews have their own API. In this case, to collect reviews, you can use the second method of collecting data - web parsing. Web parsing refers to the process of automated analysis and collection of content from the html pages of any Internet resource using special programs.

Sentiment analysis of reviews. Once the data is collected and cleaned, you can begin the process of processing it using text mining tools. To assess the author's satisfaction with the product, sentiment analysis is used in the work. Tonality or sentiment is understood as an emotional assessment of the author’s opinion in relation to the object discussed in the text.

There are three main approaches to sentiment analysis: 1) linguistic, 2) statistical, 3) combined. The linguistic approach is based on the application of rules and tone vocabularies. It is quite labor-intensive due to the need to compile tonal dictionaries, templates and build rules for determining tonality. But the main disadvantage of the approach is the impossibility of obtaining a quantitative assessment of sentiment. The statistical approach is based on supervised and unsupervised machine learning methods.

This work uses an approach based on supervised machine learning methods - a naive Bayes classifier and a support vector machine. They are quite simple to implement in software and do not require the construction of linguistic analyzers or sentiment dictionaries. Assessing the sentiment of a text can be expressed quantitatively. To apply these methods, a training set was constructed. To describe the feature space, a vector representation of the review text was used using the “bag-of-words” model. Binary vectors were considered as features - the presence or absence of a word in the review text and frequency vectors - the number of occurrences of a word in the review text. A lemmatization procedure was also used, bringing all words of the review to their initial form. More details about the machine learning methods used in this work can be found in the works.

Aspectual sentiment analysis. The work uses tone analysis of statements about aspects or aspect sentiment analysis. The sentiment aspect refers to characteristics, attributes, qualities, properties that characterize a product, for example, a phone battery or delivery time, etc. With a large number of aspects, it makes sense to combine individual aspects into aspect groups. An example of such aspect groups is presented in Figure 3. Aspect analysis of the sentiment of a review is a more complex task and consists of two stages - identifying aspects and determining the sentiment of statements about them. To solve the problem of aspect sentiment analysis, a simple and effective algorithm was developed:

First stage.

1. Extract all nouns on a set of reviews.

2. Count the frequency of use of nouns on the entire set of reviews, where is the number of uses of all words, is the number of uses of the i-th noun.

4. Sort the set of nouns in descending order.

5. Divide the set of nouns into aspect groups.

Second stage.

1. Break a lot of reviews into many offers.

2. For each sentence, perform a sentiment classification.

3. For each sentence, a check is performed: if the sentiment assessment of the sentence (negative or positive) exceeds a certain threshold and contains at least one noun from any aspect group, then the sentence is marked as a positive or negative opinion about this aspect group.

The results of sentiment analysis of reviews and aspect sentiment analysis are a set of text data, where - the text of the i-th consumer review, - the tone of the i-th review, - negative opinions about the j-th aspect group in the i-th review, - positive opinions about the j-th aspect group in the i-th review, i - review number, j - aspect group number.

Subsequent data processing using decision trees. To conduct a qualitative analysis, an original method for subsequent processing of the results of sentiment analysis based on decision trees was developed. The construction of decision trees was carried out using the C4.5 algorithm. The results obtained with its help make it possible to understand which aspect groups of products and how they influence the PM. The advantage of the developed analysis method is that it allows us to identify non-linear relationships between overall satisfaction with a product and satisfaction with its individual aspect groups. The method also allows you to identify significant aspects of the product and obtain its quantitative estimates. The method consists of the following procedures:

1. Convert a set of text data into a logical data type according to the following rules:

1.1. If , then, otherwise;

1.2. If, then, otherwise;

1.3. If, then, otherwise.

2. Construction of a decision tree in which the variable is a dependent variable on .

3. Calculation of the significance of aspect groups and interpretation of the results.

The significance of an aspect group shows how strongly the tone of a review depends on the tone of a given aspect group. Calculated after constructing a tree of classification rules. Let the number of aspect groups be equal, then the number of independent variables is equal (negative and positive statements for each aspect group). The formula for calculating the significance of the th variable will look like:

, (1)

where is the number of nodes that were split by the attribute , is the entropy of the parent node split by the attribute , is the child node for the th that was split by the attribute , is the number of examples in the corresponding nodes, is the number of child nodes for the th parent.

The assessment of consumers' UE for products is calculated using the formula:

, (2)

where is the number of positive reviews, is the number of negative reviews.

The assessment of the PP by the -aspect group of products is calculated using the formula:

, (3)

where is the number of reviews containing a positive mention of the th aspect group, is the number of reviews containing a negative mention of the th aspect group.

Experiment

The effectiveness of the developed ISSPR prototype was assessed on a data set of 635,824 reviews in Russian, dedicated to the resort and hotel business. Reviews were collected from the popular Internet resource tophotels.ru for the period 2003-2013. . The data was pre-processed (duplicates, fragments of html markup and reviews less than 30 characters long were removed) and loaded into the SQL Server 2012 database.

To classify the tone of reviews, a binary scale (negative and positive) was used. The training sample of positive and negative reviews was formed using the collected information about the author’s assessments of accommodation, food and service. The online resource tophotels.ru uses a 5-point rating scale on which food, accommodation and service are assessed. The training sample included 15,790 negative reviews with 3 and 4 total points and 15,790 positive reviews with 15 total points. The authors' assessments were not used in further data processing. The remaining 604,244 reviews were tagged using a trained classifier.

In order to build an effective sentiment classifier, the classification accuracy of machine learning algorithms and some features of their construction were assessed (Table 1). To assess the classification accuracy, the ratio of the number of correctly classified positive and negative reviews to their total number is used. Accuracy assessment was carried out on two data sets. The first set (Test No. 1) represented the formed training set. It was tested using cross-validation, dividing the data into 10 parts. The second set (Test No. 2) contained reviews with different scores and was marked manually (497 positive and 126 negative). The second set was used to monitor the accuracy of the classifiers trained on the first data set.

Table 1 - Comparison of sentiment classification accuracy

	Machine methods training
	SVM (linear kernel)	Frequency
	SVM (linear kernel)	Binary
		Binary
		Frequency
	NB (exception words)	Frequency
	NB (tagging particles "not" and "nor")	Frequency

To mark up reviews and analyze sentiment, classifier No. 6 was chosen based on the NB method, with frequency vectors as a feature space and using lemmatization techniques and tagging of negations “not” and “neither”. Using the developed algorithm, aspectual nouns were extracted from the entire set of reviews, which were divided into seven main aspectual groups (Figure 3). Next, sentences mentioning words from aspect groups were extracted and marked by tone.

A study of UE was conducted for two 5-star hotels - Hotel A (1692 reviews) and Hotel B (1300 reviews), located in the resort of Sharm el-Sheikh (63,472 reviews) in Egypt. The results of the quantitative analysis are presented below. Figure 4 shows the dynamics of the UE. Figure 5 shows satisfaction by aspect group. The “Resort” category includes reviews of all hotels in a given resort.

Figure 3 - Aspect groups of the “hotel” sentiment object

Figure 4 - Dynamics of consumer satisfaction indicators by month

Figure 5 - Consumer satisfaction by aspect groups in 2012 and 2013

To conduct a qualitative study using the developed method, decision trees were built based on reviews of all hotels in the resort and separately for hotels “A” and “B”. The extracted decision tree rules are presented in Table 2. The significances of the aspect groups are presented in Table 3.

Conducted quantitative and qualitative studies of PM for Hotel “A” made it possible to identify problematic aspect groups and identify those that have the greatest impact on PM; taking this into account, the priority of decisions made and ways to resolve them in the long and short term were determined.

Table 2 - Extracted rules using decision trees

		Key	Support1	Credibility2
Extracted rules from all resort reviews
Extracted rules from reviews of Hotel “A”
Extracted rules from reviews of Hotel B
1Support shows the proportion of reviews from the original sample that contain this rule. 2Reliability shows what proportion of reviews containing a rule have a given tone.

Table 3 - Significance of aspect groups

		Hotel "A"	Hotel B			Hotel "A"	Hotel B

Table 4 - Application of the results to develop management decisions for Hotel “A”

Problematic aspect groups	Significance	Rules with
			Train and motivate service personnel, check the quality of restaurant service, and organize entertainment events.
			Diversify the range of dishes, organize garbage collection on the beach.
			Not significant or outside the area of ​​competence.

Conclusion

The proposed concept of decision support based on the developed methodology for processing and analyzing text data allows us to automatically conduct quantitative and qualitative research on consumer satisfaction and make effective management decisions on product quality management. This concept can significantly reduce the labor intensity of consumer satisfaction research, which makes it accessible for use by a wide range of enterprises.

Based on the proposed concept, a prototype IDSPR was developed. The experiment showed the effectiveness of the approach in solving real problems of product quality management, the satisfactory accuracy of text analysis processing algorithms, and the consistency of the results obtained. IDSPR allows you to make quality management decisions based on analytical processing of text reviews from the Internet, which implicitly contain information about customer satisfaction.

Reviewers:

Chernyakhovskaya L.R., Doctor of Technical Sciences, Professor, Professor, Ufa State Aviation Technical University, Ufa.

Kartak V.M., Doctor of Physical and Mathematical Sciences, Professor, Head. Department of Applied Informatics, Bashkir State Pedagogical University named after M. Akmulla, Ufa.

Bibliographic link

Yusupova N.I., Bogdanova D.R., Boyko M.V. MATHEMATICAL SOFTWARE TO SUPPORT DECISION MAKING IN PRODUCT QUALITY MANAGEMENT BASED ON ANALYSIS OF TEXT INFORMATION // Modern problems of science and education. – 2014. – No. 3.;
URL: http://science-education.ru/ru/article/view?id=13024 (access date: 01/15/2020). We bring to your attention magazines published by the publishing house "Academy of Natural Sciences"

The quality and price of goods and services provided largely determine the success of companies. In this regard, quality management as a separate type of activity acquires special importance. Quality management refers to the planning, organization, regulation and control of activities to ensure the quality of production and management.

The object of control in quality management is the processes of ensuring the quality of products, goods, and services. There are actually three aspects of quality: conformance quality, design quality, and functional quality (the degree to which customer needs are met). When determining quality, it is necessary to take into account not only the degree of its compliance with technical conditions, but also the process of distribution of goods and after-sales service. Thus, the peculiarity of the control object lies in its complexity and the complexity of quantitative assessment.

The subject of management, i.e. decision makers - enterprise management, specialists from the marketing department, quality control department, sales and supply specialists, managers and specialists of production departments.

The main factors influencing decision making in the field of quality management include the following: ?

specialization (field of activity) of the enterprise; ?

organizational and legal form of the enterprise; ?

normative and methodological support for quality management; ?

availability of a certificate of compliance with standards (GOST, OST, 1BO); ?

established quality management practices; ?

location of the enterprise; ?

goals of the enterprise; ?

degree of centralization of management; ?

corporate culture; ?

organizational structure of the enterprise; ?

personnel qualification level; ?

quality of materials and purchased components (incoming inspection); ?

technical equipment of the enterprise, including the characteristics and capabilities of the equipment; ?

accepted procedures for completing tasks; ?

leadership style and leader experience, etc.

Solutions in the field of quality management and their features. The first step toward quality assurance is making strategic decisions about the design of a product or service to satisfy specific customer needs. There are 180 international quality standards. The fact that a company is certified according to these standards significantly increases the competitiveness of products and enterprises. in general, and also opens up the possibility of entering the international market. Therefore, the development of quality standards in written form is of utmost importance.

Within the framework of the quality management system, the following decisions are made: ?

choosing a quality improvement strategy (more careful control of manufactured products, introducing a more effective control system, purchasing higher quality raw materials, introducing more advanced technology, etc.); ?

formulation of technical conditions (quality requirements); ?

establishing qualitative and quantitative quality parameters; ?

establishing permissible deviations from specified parameters; ?

selection of control methods; ?

appointment of persons responsible for control; ?

solving behavioral problems of quality control, etc.

Features of decisions made in the field of quality management

are that they: ?

directly affect the success of the enterprise in the fierce competition for consumers, retention and conquest of new markets; ?

cover the entire cycle of creation and development of new equipment, serial production of products, sales of goods and after-sales service; ?

must take into account the socio-psychological aspects of management (including motivation); ?

focused on effective control; ?

accepted at all levels of organization management; ?

As a rule, they are reflected in documents.

Decision-making methods in the field of quality management. In table 8.14

Some methods of decision-making in the field of quality management used in the process of implementing specific tasks are presented.

Thus, decision-making in the field of quality management is a determining factor in ensuring the competitiveness of products and the enterprise as a whole, which is very important both at the tactical level and at the strategic level (at the present stage and in the future).

Table 8.14

Methods of decision-making in the field of quality management Tasks "Some methods of decision-making in the field of quality management Planning for the creation of a quality system

Quality assessment

Quality system development - forecasting methods; -

modeling methods; -

methods of multicriteria evaluation of alternatives -

methods for analyzing management decisions; -

expert assessment methods; -

statistical methods -

methods for generating alternatives; -

methods for multi-criteria evaluation of alternatives Test questions and tasks 1.

How does the manager’s field of activity influence the process of developing and making management decisions? 2.

Describe the decision-making environment in financial management. 3.

What factors influence financial management decisions? 4.

List the features of financial management solutions. 5.

What decision-making methods in the field of financial management do you know? 6.

Describe the decision-making environment in production management. 7.

What factors influence production management decisions? 8.

List the features of production management solutions. 9.

What decision-making methods in the field of production management do you know? 10.

Describe the decision-making environment in personnel management.

I. What factors influence decision-making on personnel management? 12.

List the features of HR management solutions. 13.

What decision-making methods in the field of personnel management do you know? 14.

Describe the decision-making environment in marketing. 15.

What factors influence marketing decisions? 16.

List the features of marketing management solutions. 17.

What methods of decision-making in the field of marketing management do you know? 18.

Describe the decision-making environment in quality management. 19.

What factors influence quality management decisions? 20.

List the features of quality management solutions. 21.

What methods of decision-making in the field of quality management do you know? 22.

Describe the decision-making environment in information management. 23.

What factors influence information management decisions? 24.

List the features of information management solutions. 25.

What decision-making methods in the field of information management do you know? 26.

Describe the decision-making environment in the management of intellectual assets. 27.

What factors influence decision-making on intellectual asset management? 28.

List the features of solutions for managing intellectual assets. 29.

What methods of decision-making in the field of intellectual asset management do you know? 30.

Describe the decision-making environment in strategic management. 31.

What factors influence strategic decision making? 32.

List the features of strategic decisions. 33.

What methods of making strategic decisions do you know? 34.

Describe the decision-making environment in innovation management. 35.

What factors influence decision-making on innovation management? 36.

List the features of solutions for managing innovation activities. 37.

What methods of decision-making in the field of innovation management do you know?

The best option for a decision made at one of the levels of the control system on any issue is called optimal, and the process of searching for this option is called optimization.

The complexity and interdependence of technical, organizational, socio-economic and other aspects of modern production management lead to the fact that making a management decision inevitably affects dozens and even hundreds of various factors, so intertwined with each other that it is impossible to isolate and analyze them separately using conventional analytical methods.

Many factors that determine or influence the choice of solution are, by their nature, not amenable to quantitative characteristics, while others practically cannot be measured. All this made it necessary to develop special methods that facilitate the selection of management decisions in complex technical, organizational, and economic problems (operations research methods, expert assessments, etc.).

Operations research methods are used to express optimal decisions primarily in the following areas of management: large-scale production planning; organization of production processes at enterprises; logistics; organization of transportation.

Operations research methods are based on the use of mathematical (deterministic), probabilistic models representing the process, system or type of activity being studied. Such models provide a quantitative description of the problem and serve as the basis for making management decisions when searching for the optimal option. How justified are these decisions, are they the best possible, are all the factors that determine the optimal solution taken into account and weighed, what is the criterion to determine that this solution is really the best - these are the range of questions that are of great importance for production managers, and the answer to which can be found using operations research methods. Optimization of solutions consists of a comparative study of numerical estimates of factors that cannot be assessed using conventional methods. The best possible solution for an economic system is optimal, and the best solution regarding individual elements of the system is suboptimal.

Operations research methods are designed to find solutions that would be optimal for the largest possible number of enterprises, organizations or their divisions. Quantitative methods of operations research are based on the achievements of economics, mathematics and statistics (optimal programming, queuing theory, game theory, graph theory, mathematical statistics, etc.).

Optimizing a solution is the process of searching through many factors that influence the result. The optimal solution is the most effective solution among all alternative options, selected according to some optimization criterion.

Since the optimization process is expensive, it is advisable to use it when solving strategic and tactical problems. Operational problems should be solved using, as a rule, simple, heuristic methods.

The main optimization methods are:

forecasting;

modeling (logical, physical, economic and mathematical).

In practical management, the main methods of analysis are:

comparison method;

index method;

balance method;

method of chain substitutions;

elimination method;

graphic method;

functional-cost analysis;

factor analysis;

economic and mathematical methods.

Forecasting in management refers to the process of developing forecasts itself, i.e. scientifically based judgments about the possible states of an object, the ways and timing of its modifications. Functionally, a forecast in management is presented as a pre-plan development of multivariate models for the development of a management object. The forecast is probabilistic in nature and can undergo changes under the influence of changing conditions of the external and internal environment of the organization.

The most important forecasting tasks:

developing a market forecast;

identification of economic and other trends influencing market conditions and the scale of the beneficial effect;

selection of forecasting method and time targets;

economic justification for the development or improvement of manufactured products, etc.

The main forecasting functions include:

consistency;

complexity;

continuity;

variation;

adequacy and optimality.

A model is a representation of a system object or idea in some form other than the integrity itself. It is a simplified image of a specific life (managerial) situation. In other words, the models reflect real events, circumstances, etc. in a certain way.

There are a number of reasons for using a model instead of trying to directly interact with the real world:

the complexity of the real world is such that the number of variables related to a specific problem significantly exceeds the capabilities of any person and can be comprehended by simplifying the real world using modeling);

experimentation - there are many management situations in which it is desirable to try and experimentally test alternative solutions to a problem. In addition, there are critical situations when a decision needs to be made, but experimentation cannot be done in real life;

orientation of management to the future - it is impossible to observe a phenomenon that does not yet exist and, perhaps, will never take place, as well as to conduct direct experiments. Modeling is the only systematic way to date to see future options and determine the potential consequences of alternative solutions, which allows them to be objectively compared.

Modern organizations use three basic types of models:

physical model (represents what is being studied using an enlarged or reduced description of an object or system (drawing, plan, layout);

analogue model (represents the object under study as an analogue that behaves like a real object, but does not look like one. An example of an analogue model is an organizational chart.;

mathematical model (symbolic) - symbols are used to describe the properties or characteristics of an object or event.

Building a model is a process. The main stages of this process are problem statement, construction, validation, application and updating of the model.

Statement of the problem. The first and most important stage of building a model, which can provide the correct solution to a management problem, is to state the problem. Proper use of mathematics or the computer will not be of any use unless the problem itself is accurately diagnosed. To find an acceptable or optimal solution to a problem, you need to know what it consists of. Unfortunately, sometimes huge amounts of money are spent on searching for thoughtful answers to incorrectly posed questions. Just because a manager is aware of a problem does not necessarily mean that the real problem has been identified. A leader must be able to distinguish symptoms from causes.

Building a model. The developer must determine the main purpose of the model, what output standards or information is expected to be obtained using the model to help management solve the problem facing it. It is also necessary to determine what information is required to build a model that satisfies these goals and produces the necessary information as an output.

Checking the model for reliability. One aspect of verification is to determine the extent to which the model fits the real world. A management scientist must determine whether all the essential components of the real situation are built into the model. Testing of many management models has shown that they are not perfect because they do not cover all relevant variables. Naturally, the better a model reflects the real world, the greater its potential for helping managers make good decisions. The second aspect of model testing involves determining the extent to which the information it provides actually helps management cope with the problem.

Application of the model. No model of management science can be considered successfully built until it is applied in practice. This seems obvious, but is often one of the most troubling aspects of a build.

Model update. Even if the model is successful, it will almost certainly require updating. Management may find that the form of the output is unclear or that additional data is desired. If the organization's goals change in a way that affects decision making, the model must be modified accordingly. Likewise, a change in the external environment—for example, the emergence of new customers, suppliers, or technologies—can invalidate the underlying assumptions on which the model was built.

Thus, improving the management decision-making process and, accordingly, increasing the quality of decisions made is achieved through the use of a scientific approach, models and methods of decision-making.

A management decision is the result of a manager’s activities. The effectiveness of a management decision is defined as the ratio of results to costs of its implementation. The effectiveness of a manager's activities determines the quality of management decisions.

The quality of management decisions is its characteristics that play a certain role in the management process.

In accordance with the essence and purpose, several quality characteristics are distinguished:

1. validity lies in the level of knowledge and use of actually existing laws and principles on the basis of which the enterprise develops;
2. timeliness assumes that the greater the need for a given decision at the time of its adoption, the higher the degree of its effectiveness, and therefore its quality;
3. authority increases the quality of a management decision if it is made by a person who has the appropriate rights and competence, in accordance with regulatory documents - both state and existing within the enterprise;
4. rationality is a characteristic of the quality of management decisions from the point of view of minimizing the funds invested in its development and in accordance with the costs of its implementation;
5. brevity of presentation and clarity for the performer lies in the brevity and clarity of the decision made, as well as simplicity of wording and eliminates floridity and ambiguity of wording. The next quality indicator depends on this;
6. consistency of decisions made with existing decisions and regulatory documents regulating the activities of the organization.

In the process of developing management decisions, the following factors must be taken into account:

1. characterization of the problem in terms of its complexity, novelty, degree of certainty and type;
2. the development of a problem is determined by the availability of methods, programs and skills for its implementation;
3. characteristics of information, such as volume, accessibility, reliability, relevance, etc.;
4. limited resources: time, labor, financial, material and technical, etc.;
5. organization of solution development;
6. competence, education and work experience of managers;
7. subjective factors, such as the ability of team members to work together, their cohesion, etc.;
8. information technologies with the help of which information is collected, analyzed and processed.

In the process of translating management decisions into reality, the following points are important:

1. the peculiarity of the chosen solution is characterized by its complexity, novelty, class, etc., like the problem itself;
2. organizational structure for the implementation of management decisions, i.e. the units in which it will be carried out, and the distribution of responsibility;
3. implementation deadlines;
4. competence of performers;
5. the authority of the leader among his subordinates;
6. socio-psychological factors;
7. reliability and productivity of technical controls;
8. degree of organization and control over the performer.

High quality management decisions are ensured in the case of a systematic approach to solving the problem. Science-based methods and models for their implementation should be used, which must correspond to the current situation in terms of the ratio of results to losses resulting from any decision made.

The quality of a manager's decisions undoubtedly increases with the use of modern information technologies and software developments. A significant influence on the decision made is exerted by the organization’s personnel, its qualitative composition (qualifications, age of workers, work experience, etc.), creative capabilities and the ability of its members to interact effectively (psychological compatibility).

In addition, the management decision should be as flexible as possible so that the enterprise is given the opportunity to apply methods and technologies for its implementation with minimal losses.

Despite the fact that the quality of a management decision (efficiency) is its main characteristic, determining its level is associated with a number of difficulties. It is often difficult to determine the result of a decision, especially its creative side, because, firstly, the requirements by which its quality should be assessed are rarely established, and, secondly, the result is largely related to socio-psychological factors that are not can be quantified. Most often in practice, the quality of management decisions is assessed from the point of view of their impact on profit.

The quality of a solution can be assessed based on stages development, adoption and implementation.
Assessment of solution quality at the stage production carried out when selecting possible options and choosing the final solution.

This process is carried out on the basis of objective criteria, the most common of which is the optimality criterion. But in practice, it often remains unrealized, because it does not take into account the risks associated with limited resources. To calculate this indicator in the absence of a regulatory framework, the deadline for completing work is established using statistical data, expert assessments, based on precedents, as well as the development of a network matrix.

In addition to the optimality criterion, you can use an indicator of economic efficiency, defined as the result minus costs. From this point of view, a decrease in the cost of production, an increase in its quality, profitability, etc. are assessed.

At the decision development stage, the planned economic efficiency is calculated.
At the stage acceptance decision, the final choice of the most appropriate option takes place based on economic efficiency, optimality criteria and the likelihood of its implementation. Recently, in addition to these criteria, special attention has been paid to socio-psychological factors, environmental consequences and future prospects of the organization.

At the stage implementation decisions are monitored and adjusted at the intermediate and final stages, and the deadlines set for its implementation are compared with the actual ones. The analysis is carried out on the basis of the economic effect (cost-benefit ratio).

The results of the analysis are used in the further work of management to identify further directions for the organization’s development and eliminate obstacles to achieving goals.
The main directions of development of the organization are economic, socio-psychological and organizational.

The economic direction lies in the maximum realization of the economic interests of participants in the process of developing and implementing a solution and in the use of various resources.
The socio-psychological direction involves increasing the level of professionalism of managers, forming a team on a scientific basis and developing their interest in creative activities, as well as involvement in all stages of the decision-making process.

The organizational direction concentrates its attention on increasing the level of work of management employees and improving the movement and use of means of production. Improving organizational activities leads to increased organizational effectiveness.