Efficiency Enhancement with Rule-Based Method for Credit Classification

Abstract

The efficiency enhancement with the Rule-based Method is a data mining technique to study the relationship between credit borrowers and deciding credit approval and reduce the risk of bad debt in the future. This research aims to efficiency enhancement with the Rule-based method for credit classification, which is the credit types data, numeric, and nominal used for the category from the cooperative savings database by using the Gain Ratio as a measurement unit of the sampling (Entropy) and filter to select important variables. Therefore, the researcher uses the K-fold cross-validation method by dividing the data to perform the test into equal K-part numbers into training and testing data sets. Then Rule-based approach of data mining techniques in WEKA software version 3.9.4 viz Decision Table, RIPPER (JRip), OneR, and Partial Rule (PART) to efficiency enhancement of the model for credit classification to get more accurate and reliable by measuring the efficiency of the model with Recall, Precision, and F-measure. The results of the research can be found that both the Gain Ratio and the outlier data filter can make the efficiency of the model with the Rule-based method using the Partial Rule to get the highest Recall value of 4.1%, the highest Precision value of 4.0%, and highest F-measure value by 5.4%. Besides, the Partial Rule can make the model's efficiency for credit classification get a Recall of 86.1%, Precision of 85.9%, and F-measure of 85.6%. Thus, all values were more efficient than the Decision Table, JRip, and OneR.