The design of a PC based financial credit evaluation system involving an artificial neural network for the evaluation of industrial manufacturers
The Financial Loan Evaluation System (FLES) is based on a DOS operation system program to evaluate industrial manufacturers' financial situations. The FLES utilizes an artificial neural network (backpropagation) and evaluates the firm's financial loan proposal, based primarily on the financial ratio analysis method. Since the system is able to learn from past evaluation examples, it performs somewhat like an experienced human expert. Thus, unlike other expert system applications, the FLES allows a user to rebuild the system with less development time and cost when conditions and/or rules have changed. In this thesis, FLES is trained by fifty loan evaluation examples. In the training examples, the inputs and target outputs are required to build a backpropagation network. The input data include financial ratios and finance-related information, while the output is a percentage of the proposed loan amount granted. The desired outputs are generated by the constant pair-wise comparison method because they are not provided by library sources as are the financial ratios. The constant pair-wise comparison method determines what decision making criteria are the most and the least important by calculating for each a numerical weight. The FLES shows adequate responses in more than 75% of the loan evaluations when it is evaluated with fifty new examples. FLES is relatively easy to use and is capable of providing an answer to a user in approximately 30 to 60 minutes.
School Location:USA - Ohio
Source Type:Master's Thesis
Keywords:pc financial credit evaluation system artificial neural network industrial manufacturers
Date of Publication:01/01/1994