Automation of Financial Loaning System: Accuracy Comparison of ID3 & C4.5 on Financial Data

摘要

Classification Algorithms play significant role by building model from existing data that can be used to predict the behavior of new data. This paper considers the algorithmic efficiency and accuracy of classification algorithms for financial data where a high need for accurate results exists. The new concept of data-view is also introduced in the paper accompanied with its effect on accuracy levels. The main idea is to examine the financial dataset under different classification algorithms with different mining attribute selections. In this way we can evaluate the most suited algorithm and combination of attributes for building the final classification model for financial data, which will increase the prediction accuracy and decision of loaning a client based on his past transactions, will be more reliably made. It can be very helpful for a financial institution to develop an automated loaning system with least chance of error and fraud.