DonNTU   Masters' portal

Abstract

Содержание

Introduction

Changes, what be going on in the economy of Ukraine, suppose substantial changes in the mutual relations of banks with the subjects of management – enterprises, organizations, other jars. Jars as commercial organizations the basic operations of which it is been crediting are a calculation, deposit, cashdesk and other operations, carry the most various risks during their leadthrough: nevozvrat of the given out credit, non-payment of percents on a loan, risks are a calculation, currency, interest rates etc. In the everyday activity of jar constantly run into a risk to lose part of the facilities, this risk is inevitable and, but it is always necessary to aim to decrease him – an income and viability of bank depend straight on it. One of basic bank risks is a credit risk.

1. Purpose and Objectives

The aim of this work is to develop techniques and algorithms for risk assessment based on mining semi-structured data on the credit behavior of the borrower. To achieve this goal requires the following tasks:

  1. Analysis of the characteristics of risk management of banks in lending to borrowers.
  2. Analysis and selection of models to predict credit risk based on credit profiles of borrowers' behavior.
  3. Analysis of existing methodologies for assessing candidates for borrowers and their impact on the assessment of credit risk.
  4. Development of a technique of building a knowledge model that reflects the risk assessment based on the model of subjective credit behavior of the borrower.
  5. Development of a modified method of determining a mortgage through riskassessment based on the solvency and credit behavior of the subjective.

The object of this study is lending activities of the bank and credit risk management.

The subject of research is the algorithmic support risk prediction based on semi-structured data on the credit behavior of borrowers.

2.Actuality of theme of work

The loan is one of the most difficult economic categories, the research entity which holds an important place in the work of national and international scientists. The rapid credit growth, which is accompanied by increased risk compared to other types of banking activities and lower yields, the need exists for new approaches, methods and techniques to the management of the bank's loan portfolio, development of an effective mechanism for the loan process and use it in practice. The main component of improving the effectiveness of the loan – the evaluation of the risk of this type of service. Credit risk management is an essential part of the strategy and tactics of survival and development of any bank. Lending has always been and remains a priority economic function of banks.

Borrowing money from credit institutions at the present time is a common way to expand their financial capabilities. For explanation of credit borrowers are divided into bona fide – return loan funds, and fraud – using the money and are not going to return them. Credit scoring [1] is a system for assessing the borrower bank, the result of this assessment is a decision on the loan application if the borrower accumulates a certain number of points, then he is given credit.

The introduction of automated systems for evaluating the credit of the borrower, as the most effective method of preventing fraud on the basis of existing evidence does not lead to the desired results, as well as in cases of use of the conclusions of a credit bureau, and may not be fundamental in making a decision on a loan because the level of defaults is high enough. The use of false personal data which do not reflect the real intention of borrowers, contributes not only wrong decision made in the scoring, but also leads to an accumulation of conflicting statistics.

People's attitudes toward money is systemic in nature, different kinds of relationships are interrelated, and each uniquely special relationship manifested various other types of relationships. Consequently, the psychological evaluation to the action with the money and getting credit can be a tool for assessing integrity of the borrower loan. To measure attitudes toward money and transactions with them are being developed Psychosemantic various techniques [2]. Moreover, these techniques are also used to assess borrowers motifs [3]. Therefore important to use the capabilities of modern psychological methods to solve the problems of scoring. The development of software for risk assessment based on financial analysis involved in domestic and foreign companies: R-Style Soft lab, «Analytical Technology Business» – dm Score, SAS-Credit Scoring for Banking, Camel, STATISTICA, EQUIS, Hyper Logic.

The analysis of existing products shows that they do not use psychological models, the current level of development of the theoretical, methodological and algorithmic issues of risk analysis of credit transactions require improvement. Thus, the development of methods and algorithms for assessing bank risk models based on the borrower's credit behavior is relevant.

3. Prospective scientific novelty

  1. A modified algorithm for extracting and structuring knowledge to determine the feature space to reflect the borrower's credit behavior (CP) using a model based on the typology of a product fault rule base (BP) tree structure.
  2. The method of constructing a knowledge model that reflects the risk assessment based on the model of subjective credit behavior of the borrower.
  3. The algorithm for determining a mortgage on the basis of risk assessments based on the solvency and credit behavior of the subjective.

4. Planned practical results

The practical significance of the study is to improve the effectiveness of risk assessment as the main component of the lending system by improving the efficiency of the classification process faults as well as reducing the amount of data requiring documentary evidence of financial viability of the applicant. Developed algorithmic software can be used in the development of software tools to support lending services.

In writing this essay master's work is in an unfinished state. The final completion of the work will be in December 2013. Full text of the work and materials on the topic can be obtained from the author or his manager after that date.

References

  1. Економічний ризик і методи його вимірювання: Навчальний посібник. Київ: Центр навчальної літератури, 2003. 188 с.
  2. Петренко В. Ф. Основы психосемантики. – М.: Московский университет, 1997. – 400 с.
  3. Семенов М.Ю. Представленность денег в образе мира/ Семенов М.Ю.// Сибирский психологический журнал. — 2000 а. — Вып. 12.— С. 66-67.
  4. Altman, E. Credit Risk Measurement: Developments over the Last 20 Years [Text] / E. Altman, A. Saunders // Journal of Banking and Fi-nance. — 1998. — №21. — P. 1721-1742.
  5. Астрелина, В. Новое Базельское соглашение [Текст] / В. Астрелина, А. Мирошниченко // Управление финансовыми рисками. — 2005. — № 1. — С. 2-9.
  6. HR1-Кредит. Система тестирования и оценки клиента банка для выявления риска невозврата кредита. Интернет-ресурс. — Режим доступа: http://www.akvilona.ru/brands/nemesysco/description/mod574201 31403.htm
  7. Энциклопедия финансового риск-менеджмента / Под ред. А. А. Лобанова и А.В. Чугунова. — М: Альпина Паблишер, 2003. — 786с.
  8. Кузнєцова Н.В., Бідюк П.І. Порівняльний аналіз характеристик моделей оцінювання ризиків кредитування// Наукові вісті НТУУ “КПІ”. — 2010. — №1. — С. 42-52.
  9. Шепелева М.В. Разработка экспертной системы оценки рисков при кредитовании физических лиц / автореферат к магистерской работе // http://www.masters.donntu.ru/2006/kita/shepeleva/diss/index.html
  10. Таценко А.Г. Разработка модели оценки риска при выдаче кредита физическим лицам/ автореферат к магистерской работе // http://www. http://masters.donntu.ru/2008/fvti/tacenko/diss/index.htm
  11. Бурцева Е.Ю. Разработка компьютеризированной подсистемы оптимизации управления потребительским кредитованием коммерческими банками / автореферат к магистерской работе // http://www. http://masters.donntu.ru/2008/kita/byrtseva/diss/index.htm
  12. Kiss, F. Credit scoring processes from a knowledge Management perspective / F. Kiss // Periodica Polytechnica. Ser. Soc. Man. Scl. — 2003. — Vol. 11, № 1. — P. 95-110.
  13. S.Murthy. Automatic construction of decision trees from data: A Multi-disciplinary survey.1997.