Intelligent Integrated Systems of Automated Design Laboratory ( IISAD )

We are presenting the new techniques for quantitative assessment of risk in finances and economics based on logical and probabilistic methods (LPM). The risk models in these area have logical connection OR, AND, NOT and groups of incompatible events. LP-techniques have almost twice more accuracy in recognition of good and bad credits than used techniques based on LDA, QDA, CART, NN methods.

The demonstrated software based on LP - technique is developed. It's demonstrating on the dataset of 1000 credits of individuals how to decide the following tasks:

1) to calculate a risk of the credit (probability of default), to classify credit on good and bad, to set acceptable risk (cut off value) and average bank risk, to determine contributions of the credit's characteristics in risk;

2) to perform the analysis of bank activity, having calculated average risks and contributions of characteristics of all credits of bank and to identificate mistakes, which bank makes more often; to nominate well-founded acceptable risk and the price for risk;

3) to perform identification (training) of risk model (to calculate the probabilities of initiating events) using bank statistical data about credit issues.

The demonstrated software consists of the packed files RISK.ZIP, including files:

After unpacking of ZIP-file you should run the executing file RISK.EXE, which works in MS DOS. The software works in dialogue with an user.

We have also demo-software for Windows 98, XP.

We ask you send remarks about the demonstrated software to E_mail address: risk@sapr.ipme.ru