A Comparison of Normalization Data Transformation Efficiency Affecting with Bank Customer Credit Data Classification using Data Mining Techniques

Abstract

The objectives of this research 1) to compare the effects of data transformation using 3 methods namely Min-Max, Z-score and Mean 2) to study the efficiency of data classification of 3 techniques namely KNearest Neighbor, Decision Tree and Neural Network. This research studied credit approval data of bank customers. The results showed that the Knn-MinMax, Knn-Mean and Knn-Z-score techniques yielded 80.63% classification efficiency, while the second most efficient technique, DT-Mean, yielded the second 80.53% classification efficiency. Next, DT-MinMax and DT-Z-scores had provided a classification efficiency of 80.43%. Finaly, the NN-MinMax, NN-Mean, NN-Z-score techniques provide a classification efficiency of 77.45%. Therefore, the techniques of normalization data transformation into Min-Max, Z-score and Mean. They have highest effect on the K-Nearest Neighbor classification technique because they provide the highest classification efficiency. The second most effective technique for normalization data transformation was Decision Tree technique and the last one was the Neural Network technique.