Труды Кубанского государственного аграрного университета


<<<  Back

2016, № 58

UDC: 004.032.26:336.77
GSNTI: 06.73.55

Neural network estimation of physical persons’ credit status in the credit organizations

Possibilities of artificial neural networks are studied in the article, the area of their use is expanded, creation and realization of the software product allowing automation of the assessment process of solvency of natural persons by banks or other credit organizations on the basis of use of neural networks is described. The analysis of literature showed that the problem of solvency assessment of a client was solved and is solved by some authors both for legal, and for natural persons. However the description of the decision with application of particular bank data, process of operation of a certain neuro-simulator was found in not a single source. This is the scientific novelty and specifics of the work. Relevance of work is conditioned by the need of further mathematical and instrumental credit risks management tools improvement aimed at optimal distribution of financial resources of banks. The bank borrowers qualifier creation algorithm or other credit organization on the basis of neural networks is suggested - the difficult process, including the following stages: database compilation from examples, splitting them on two sets: training and testing; preliminary data processing (a normalization, standardization, etc.), transformation for giving on an entrance; designing (a choice of topology of a neural network), training and network quality assessment on the basis of the confirming set; preservation of a network option which provides the required classification accuracy (small quantity of incorrectly recognizable examples); practical use of a network for the solution of borrowers classification problem. All stages of the specified algorithm are worked in details, described and presented by screenshots of the program work in an available form. Thus, practical use of the Statistika 8 Neural Networks neuro-simulator for the solution of the problem of borrowers’ solvency assessment - natural persons - is shown, the competitiveness of this affordable software product with foreign packages bought by banks and the credit organizations is demonstrated.
Keywords: Neural networks, problem of classification, problem of decision-making, scoring, solvency assessment

References:

  1. Абричкина, Г. Б. Инструментальные методы управления кредитными рисками регионального банка / Г. Б. Абричкина. - Дис.. канд. экон. наук. - Воронеж, 2004. - 156 c.
  2. Бэстенс, Д.-Э. Нейронные сети и финансовые рынки. Принятие решений в торговых операциях / Д.-Э. Бэстенс, В. Ван Ден Берг -М., Вуд Д. - М., 1997.
  3. Боровиков, В. П. Нейронные сети. Методология и технологии современного анализа данных / В. П. Боровиков. - 2 изд. - Москва, 2008.
  4. Боровиков, В. П. Популярное введение в программу STATISTICA / В. П. Боровиков. - М., 1998.
  5. Ежов, А. А. Нейрокомпьютинг и его приложения в экономике и бизнесе / А. А. Ежов, С. А. Шумский. - М., 1998.
  6. Кричевский, М. Л. Нейронные сети в задачах риск-менеджмента / М. Л. Кричевский. - Научн. сессия проф.-преп. состава СПбГУЭФ. - Сб. докл. СПб, 2002.
  7. Корниенко, С. Л. Оценка кредитоспособности заемщика в процессе управления кредитным риском / С. Л. Корниенко. - Дис.. канд. экон. наук. - М., 2003. - 230 c.
  8. Просалова, В. С. Оценка банками кредитоспособности юридических лиц в условиях развития нейросетевых экспертных систем: На примере Приморского края / В. С. Просалова. - Дис.. канд. экон. наук. - Владивосток, 2006. - 186 с.
  9. Райзберг, Б. А. Современный экономический словарь / Б. А. Райзберг, Л. Ш. Лозовский, Е. Б. Стародубцева. - 5-е изд., перераб. и доп. М., 2006.
  10. Рындина, С. В. Математические модели application-скоринга / С. В. Рындина // Проблемы информатики в образовании, управлении, экономике и технике. - Сб. статей Всерос. научно-техн. конф. - Пенза, 2008. - С. 32-35.

Authors:

  1. Sokolova Irina Vladimirovna, Cand. PED. Sciences, associate Professor, Department of mathematics, Kuban State Agrarian University.