fuzzy neural network
past due debt
L.R. Chernyakhovskaya, K.S. Gendel, E.R. Gareeva
The article is devoted to the development of recommendations to improve the lending process, based on the algorithm makes the analytic dependence of income on the amount of the credit institution issuing different types of loan programs of the commercial bank.
Proposed to solve the problem of forecasting using a hybrid fuzzy logic system that integrates the subjective knowledge of experts and the results of the neural network analysis. The basic idea underlying the models of hybrid fuzzy inference systems, is to use the existing sample data to determine the parameters of membership functions that best meet both the expert knowledge, expressed in the form of rules, and the objective data of the process of lending. In this case, to find the parameters of membership functions using well known procedures for training neural networks.
Design a neuro-fuzzy network, realizing a dependency analysis of the bank issuing the credit programs of the revenue.
Then studied the resulting neuro-fuzzy network. After training, neuro-fuzzy network, visualize the surface of the model of fuzzy inference income depending on the number of different loan programs issuing bank. The analysis of the surface, which are formed on the basis of the conclusions and predictions.
- Guzairov M.B., CHernjakhovskaja L.R., Gerasimova I.B., Nugaeva K.R. Podderzhka prinjatija reshenijj po upravleniju kachestvom obrazovatelnogo processa s ispolzovaniem nejjronnykh setejj // Nejjrokompjutery: razrabotka, primenenie. 2007. Vyp. 10. S. 62-65.
- CHernjakhovskaja L.R., Nizamutdinov M.M. Analiz processov upravlenija v kriticheskikh situacijakh na osnove klassifikacii s ispolzovaniem nejjronnykh setejj // Nejjrokompjutery: razrabotka, primenenie. 2001. S. 73-76.
- Ljuu JU-D. Metody i algoritmy finansovojj matematiki. Laboratorija znanijj. 2009.
- DjakonovV.P. MATLAB R2006/2007/2008 + Simulink 5/6/7. Osnovy primenenija. Izd 2-e., pererab. i dop. M.: SOLON-PRESS. 2008.