Comparative evaluation of Multi Layer Perceptrons to hybrid Multi Layer Percetrons with Multicriteria Hierarchical Discrimination and Logistic Regression in Corporate Financial Analysis

摘要

The net present value of corporations is reflected in the stock price, leading portfolio mangers to maximize their assets, securing stockholders interests. Extended accounting data produced by accounting reports, and financial markets include lucrative hidden information. Econometrics, Neural Networks and Multicriteria Analysis classify companies describing their economic robustness. In a detailed comparison Multi Layer Perceptron classification is compared to neuro-genetic hybrid of Multi Layer Perceptron, Logistic Regressions and Multigroup Hierarchical Discrimination classifications to determine efficient methods in Financial Analysis. Simple Logistic and Logistic Model Trees methods had a fine performance.