2014. No. 1 (27)

Maxim Khivintcev - Post-Graduate Student, Department of Business Analytics, Faculty of Business Informatics, National Research University Higher School of Economics.
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: mkhivintsev@hse.ru

Andranik Akopov - Professor, Department of Business Analytics, Faculty of Business Informatics, National Research University Higher School of Economics.
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: aakopov@hse.ru

     The article presents a new approach to applying a multi-agent genetic algorithm (MAGAMO) for search of optimum strategic and operational solutions in large-scale simulation models.
     The purpose of the paper is to develop a simulation model of a referential Internet shop on the basis of the system dynamics methods and to apply the multi-agent genetic algorithm (MAGAMO) for solution a multi-criteria optimizing problem of the strategic and operational control parameters related to the class of large-scale problems. An imitation modeling system Powersim Studio is used for implementing the mathematical model of referential internet shop
     The research is focused on the large-scale multi-criteria optimizing problems run in simulation systems.
     For the solution of such problems, a multi-agent genetic algorithm (MAGAMO) is offered. The feature of this algorithm is the distribution of a set of the system operating parameters between agents on the basis of the preliminary cluster analysis. Each agent represents independent genetic algorithm with its own evolution of the decisions, corresponding to the preset control parameters. Information exchange between the agents functioning in parallel processes is carried out through divided memory of system (a multidimensional database). Here, the central process is responsible for selecting solutions of the highest rank of Pareto. Using a specialized software of Pareto Front Viewer visualization, Pareto’s front is provided.
     The developed simulation model is integrated with algorithm of MAGAMO, system of visualization of Pareto front and a multidimensional database.
     The results of numerical experiments, which have been carried out on real data of the internet shop, have demonstrated high efficiency of the developed multi-agent genetic algorithm for search of optimum solutions in systems of imitating modeling of big dimension.

Alexander Morgunov - Associate Professor, Department of Corporate Information Systems, Faculty of Business Informatics, National Research University Higher School of Economics. 
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: amorgunov@hse.ru

     Implementation of information systems in a large company with a wide network of branches is usually associated with a number of difficulties of organizational and technical nature. Implementation projects are time consuming not always lead to desired results.
     The paper discusses the problems related to implementation of the information system of automation of technological processes and workflow of subscription for periodical printed publications in branches of Federal State Unitary Enterprise (FSUE) «Russian Post». General performance of the business process automation before the implementation is estimated, organizational and technical problems that arise during implementation of a new information system are examined. Such problems are associated with continuity of the technological process, impossibility of performing operation with «empty database», problems of personnel development, unification of business processes, as well as large quantity of implementation objects (more than 42 000). As a result of the study the advantages of implementation of the unified information system, which may be gained by post operators are analyzed.
     The article’s matters can be of interest to experts studying the implementation of information system.

Dmitry Isaev - Associate Professor, Department of Business Analytics, Faculty of Business Informatics, National Research University Higher School of Economics. 
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: disaev@hse.ru

     The paper focuses on estimating maturity of performance management systems, which are considered according to one of the possible treatments, as systems for information support of corporate governance and strategic management. Such systems are focused into the tasks of gathering, storage, analytical processing and presentation of information, which is critical for organizations’ information transparency and strategic decision making performed by external and internal stakeholders. The purpose of this paper is to advance an approach to evaluating maturity of such systems. For this purpose, we have considered existing approaches to maturity evaluation for management and information systems, formulated general principles of maturity evaluation, and developed methodological recommendations in the field of performance management systems maturity evaluation.
     The proposed approach to evaluating performance management system maturity relies on its hierarchical conceptual model that includes such elements as functional blocks, functional modules and analytical functions. In this case a ‘bottom-up’ principle is applicable: evaluation of higher level elements maturity (up to the system as a whole) is performed relying on estimates of subordinated lower level elements. At that, every element is estimated from viewpoints of data processing methods and processes, information systems, personnel, as well as integration with complementary elements, data quality, effectiveness and governance. Advisability of maturity evaluation in dynamics, as well as comparison with certain target levels is also justified.

Konstantin Nagaev - Senior Researcher, Center for Information Analysis Applications, Institute for Statistical Studies and Economics of Knowledge, National Research University Higher School of Economics. 
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: knagaev@hse.ru

Elena Kurbatova - Junior Researcher, Center for Information Analysis Applications, Institute for Statistical Studies and Economics of Knowledge, National Research University Higher School of Economics. 
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: ekurbatova@hse.ru

     The paper discusses the problems related to data collection, processing and expert opinions evaluating in terms of a specified knowledge domain. Methods for expert data processing based on computer systems are presented. The results are used in foresight projects, long-term forecasting and in technology road mapping. The main advantage of proposed methods is the consideration of expert competence degrees. Additionally, expert datasets are adapted to aspects of the knowledge domain in accordance with a customer’s interest area.
     The proposed method provides obtaining subsets of expert data for a required version of expert polls, or expert data values effective on a specified date. An additional driver of expert data model flexibility is metadata for expert competence identification in different aspects: science, technology, business and governance. The metadata is designed to run evaluation of various analytic indicators for foresight investigations, for example, for weak signals and wild cards. The method gives accent to the opinion of a high-competence expert which may diverge significantly from the major opinion. This very property of the method is extremely important for weak signal retrieval. Wild cards are founded with an estimation of realization probability of a roadmap element with high influence rate in an interest area.
     Another area of the method applications is the development of different forecasts of knowledge domain’s evolution taking into account the  importance of properties in special contexts (economic, political, ecological and technological). Joint analysis of these data and experts’ competence data allows generating improved specific questionnaires for distinct expert groups automatically.
     The methods described here have been implemented in the software system “Interactive roadmap with feedback link” and tested in two pilot projects: “Catalytic cracking” and “Biotechnologies medical”.

Sergey Yampolsky - Associate Professor, Department of Business Analytics, Faculty of Business Informatics, National Research University Higher School of Economics. 
Address: 20, Myasnitskaya str., Moscow, 101000, Russian Federation.
E-mail: syampolsky@hse.ru

Anatoly Shalamov - Researcher, Department of Statistical Problems of Informatics and Management, Institute of Informatics Problems of the Russian Academy of Sciences.
Address: 44, Vavilova str., Moscow, 119333, Russian Federation.
E-mail: a-shal5@yandex.ru

     The article covers issues of effectiveness of working capital management of a trading enterprise on the basis of the automated planning. This approach gives the possibility to use the classical methods of optimization of dynamic systems to determine the main parameters of the economic policy of the commercial enterprise for providing the best strategy for development. Using the proposed approach, algorithms we can be created for evaluating both managed and unmanaged risk and also for finding reasonable solutions for preventing them.
     When presenting the material, operational environment of the commercial enterprise is describing in traditional terms of trade and financial market, allowing its use everywhere where the information system "1C" is run, in the form of incremental software complex for forecasting and optimization.
     The developed mathematical model allows to solve the tasks of easing the selection and justification of decisions for the managers of a shopping enterprise, including these: prediction of time functions defining mathematical expectations of assets and liabilities, definition of share capital providing strategic objectives, identification of primary commodity-monetary policy, evaluation of efficiency of variants of produced plans.
     Results of forecasting the dynamics of the situation of circulating assets are presented in an integrated graphical form, which provides an opportunity to see the full picture of the forthcoming state of current assets of a trade organization, and the main participants of the trade and economic activity in a given time interval.

Alexander D’yakonov - Professor, Department of Mathematical Methods of Forecasting, Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University; Senior Researcher, Dorodnitsyn Computing Center, Russian Academy of Sciences  
Address: 1, build. 52, Lomonosov Moscow State University, Leninskie Gory, GSP-1, Moscow, 119991, Russian Federation.
E-mail: djakonov@mail.ru

     We consider two tasks in describing a supermarkets clients’ behavior: prediction of a client's next visit date and prediction of his/her spends. The first problem is equal to estimating visit probability, and the second – to estimating density for visitor spends. To solve these problems, we propose using weighed methods: real non-negative value (weight) is assigned to every event. Weights allow considering additional information, for example history (earlier visits have smaller weights). We consider several weighted schemes (methods of assigning weights to events) and weights optimization (performance optimization by changing weight parameters). The paper shows that weighted methods don't lead to overfitting, i.e. learning on a training set doesn't decrease performance on an independent test set. We can see, that assemblers of different methods can increase performance (we consider linear combination of probabilities estimated by different methods). All experiments are made on real data of large International competition on data mining. The last span of statistics does not contain holidays, which allows concentrating only on statistical methods of problems solving while solving these tasks. Besides, we also considered construction of algorithm to solve the problems (next visit date and spends prediction) simultaneously. It can be seen that the problems not always can be solved independently. We propose a function to estimate solutions of both problems and optimization method for this function.