Rozetka UA

Analysis and Forecast

Dmitry Gorchilin

dmitrygorchilin@gmail.com
Work- use analytical methodology UTS (general systems theory) to analyze business processes and predict their condition.

Formulation of the problem: the task of this work is to develop a universal method of analysis of business processes and environmental factors, the identification of relationships and regularities, the study and analysis of these factors in order to predict the state of the system in the future. In addition, the possibility of false detection of the source data, the identification of conditions prior to a qualitative change in the system. Work is carried out with a minimum expenditure of time and resources.

The theoretical justification: human activity, as a rule, is systemic. Systems theory treats processes occurring in the world around us as the interaction between systems of different types and levels (economic, technological, biological, social, political). Interoperability can learn to identify patterns such interactions to predict the state of the system (or systems) in the future. Furthermore, on the basis of expert judgment regarding individual parameters of the system at some point in time we can calculate the value of other parameters. This knowledge will allow the best realize the benefits of avoiding possible loss. For a man control system (no matter what, whether the company, a nuclear reactor or a car) vision of the future state of the system and environmental factors, see the result of their actions on the impact of this condition- is extremely important. This vision is realized in different ways. First of all, the knowledge and experience of a competent professional. In addition, in some cases, expert assistance. But the possibility of human Disabled framework. To go beyond that used different methods, up to the horoscopes. In this case, the mathematical procedure is technically easy realizable, easily verifiable.

The problem is systemic approach is this: the real world is a complex stochastic system. That is, a logical analysis of this system is impossible because of the huge size. Parameters of system status often exposed to random and less predictable factors. That is, it is not clear that the proposed study, which is not known how it works. Is it possible? Yeah maybe. First of all, in a stable state among the various logic systems typically monotonous. Interoperability is a logical, predictable. Without knowing all the objects in the system and their relationships with each other, we can fix some of the parameters. Roughly speaking, we have a pile of black boxes, they are connected by wires. If there is an opportunity to place these wires instruments (instruments as much as possible) — we will be able to assess the condition of the interaction of these black boxes in general.

Exactly knowing the physical phenomena washed away, without providing exact structure of objects and their relationships, we can use an array of information about the system. If we have data on the operation of the system over a long period of time, mathematical methods already possible to establish the dependence of some parameters on the other, to assess What are the parameters for the system are the control actions. Several stands out the will of individuals. The role of personality in history is enormous. At the same time, the actions of individuals usually subordinate logic. If we examine the car as a system, when you start climbing the mountain road the driver, as a rule, the pressure on the gas pedal. Accordingly, it can be assumed that the increase engine power, increase fuel consumption. Will people aspirations leaders rarely defined priduryu valiant. In most cases the basis for political decision-making is the logic of calculation, the achievement of certain goals. Long-term monitoring of the system allows us to understand how a person will act in a certain traditional for the man of the situation.

That is, first, the selection of source data relating to the subject area. Any given does not happen much. In the end, it all comes down to the table, such as the status of certain variables on a certain date. To identify some values ​​depending on other methods exist, such as correlation. There are methods of approximation (usually a polynomial). But these methods are imperfect. The result of research is that of yes, this column is affected by this. Moreover, the interpolation can be relatively accurate, but extrapolation of the warranty of any kind does not.

The proposed solution is to move the family empirical formulas relating values ​​column of the table. It sounds simple, but for such a huge amount of computing breeding operations, a great job with databases. Synthetic methods, numerical methods, plus research functions. Specialization in two dimensions is a sort of Himalayan mountains, multi-dimensional study of complex orders of magnitude. This work is complex, many have considered J Fortunately, modern computers have the necessary speed, they are capable of storing vast amounts of data. More togo despite the volume of calculations the cost of the work of analysts and programmers still higher.

As a kind of concept demonstrator program written in Java. For a limited time (minutes), she reveals empirical formulas relating some initial array of test data. The problem demonstrator that made the choice of several formulas which form the original array. In reality, the data will need to accumulate on the SQL server, rewriting logic for SQL. Although the demonstration program is able to handle an array of 10-15 million, double the numbers even out of the box Java. In principle, there is no technical limitation on the number of formulas evaluated for applicability. Moreover, it is possible to automatically generate these formulas, with the help of a special program. Automatically same and check them.

What if some line does not match the calculation formula? It’s fine! On this line you need to pay close attention. The data may be distorted by deliberately. Here’s what ugodno- yes even malice or an attempt to hide phony statistics facts. The second option, we recorded a transitional process in which the interaction of the parts of the system changes with respect to a stable state. Understanding the essence of this protsessa- question to the experts. Whether it’s the beginning of the boiling water in the kettle, the bankruptcy of a firm or a revolution. The method does not explain the essence of the state, it only marks the red flag is the place.

Why is it all necessary? At one time he served at the facility in Sofrino. Notable was a place even came to see Gorbachev. We drove next time tripper-General, showed the Soviet supercomputers IAC Elbrus-2M, pocketed huge room buzzing and flashing cabinets, wall panels with large poisonous green alphanumeric indicators, even the Soviet terminals, with the body of good-quality sheet steel. In general there was a question that he could not articulate, in terms of why it’s necessary, why this abyss of money spent? Some savvy senior lieutenant found: look here nazhal- button and immediately see that the enemy has started! All were satisfied, including General. Lying is not the future I do not have a button, pressing that can predict the dollar in the coming year, as well as print a schedule of wars and revolutions in the next five years. But I have a technique using which, relying on predictable parameters state of a system can assume the next state of the system, the value of other parameters. We can not say that the procedure is absolutely reliable. But if the probability forecast of over 50% — this is good, this is more a statistical error. To make predictions about the reliability of their predictions is a thankless task, but it would be enough to take advantage of them professionals and executives have not regretted about staff time.

How exactly predict the future using this technique? It all starts with an expert. Expert formulates some initial condition of the state. Part of the data may be used without modification. For example, the population of a certain region are unlikely to change much in a short time. Further, the expert suggests the values ​​of some variables. Let’s become aware of the additional issue. Thus, the initial conditions are formulated. If their adequacy is possible to calculate some values ​​(for example, the dollar, the sales volume of a product).

The technique allows the subject to answer questions that arise in business, politics or technology. «What will be the volume of sales of this product in the ..?», «What amount of deposits will be prematurely terminated if …?». In principle, the technique is not much difference that schitat- though the parameters of a complex technological equipment, even though the financial performance of organization, even political activity in a country. All that can be summarized in a table, all presented figures can shortchange. Shortchange, find patterns and relationships. In fact, in some cases, even the technique is not very important what kind of numbers. They may be made without comment, which drastically reduces the possibility of the disclosure of proprietary information. Furthermore, in some cases it will be possible to draw the attention of experts there is some problem. A imenno- by the calculated volume of sales of certain goods or services, a revenue point of sale has to be more. That is, there is a problem that should be identified and eliminated.

Huge possibilities of application of the program in the banking activities. Starting from the calculation of projected balances on correspondent accounts with the NBU (the main work positioner) to the motion for a separate type of accounts as on the chart of accounts and the analytical characteristics, from cash balances up to the demand for certain types of banking services.

The further development of methodology: The proposed technique has one major flaw, it works with statistical cuts. This is a photograph. Yes, pictures can say that in the current system is somehow not very good, but the systems theory involves the study of the dynamics of the system. Dynamic sostavlyayuschaya- is the use of the procedure in the other rows of the table. Something happened a week ago triggered the current changes, the system usually inertia. Understand what you need to consider is, understanding how imenno- too, but we have to work on it.

Second scenario: an expert system. Fixing some qualitative changes in the system, you can understand what changes the parameters lead to them. These states should occur often, it should be a big database. Then it is possible to formulate the so-called heuristics, in this case, the logical rules. In principle, you can create a self-learning system.

The third option, in fact, the methodology of systems theory. Decomposition, the study of components and subsystems. Most of the analytical work can be done based on the existing methodology, it is already providing the necessary tools. Moreover, even knowing of the existence of the important elements of the system, it will be possible to identify them by the fact of influence on the other elements. For example, the black hole at the center of our galaxy was able to identify, after one of the stars made a sharp turn in the space around a point in space, an invisible object.

Timeline for implementation: preliminary estimates for the selection of data, preparation of hardware and software debugging will need to spend about a month. A month later, we can talk about the first primenimymh in practice results.

About the Author: Gorchilin Dmitry Ilich, 45 years old, higher education, 1994 NMetAU, with honors.

1997-2000 year: Metallurgical Plant named after Petrovsky, engineer of production department, marketing department engineer, senior engineer of the department of marketing, head of the engineering bureau.

2000-2015 year: a big bank system to support the bank’s trading day

Now, fly in the ointment: Here there is a sign, a miracle Start program, give a forecast of the dollar over the next six months! Alas, so it is not working. Even the selection of the function is a long and laborious work. Initial parameters znacheniya- very carefully. But the methodology allows to solve the problems of the physical device, which we can have no conception. It’s difficult, it’s interesting.