Prediction of Credit-Card Defaulters: A Comparative Study on Performance of Classifiers

Data mining is an emerging area of research that aims at extracting meaningful patterns from available data. This paper highlights the significance of classification in predicting new trends from voluminous data. Performance analysis of various data mining algorithm viz. BayesNet, Meta-Stacking, Naïve Bayes, Random Forest, SMO and ZeroR in predicting creditcard defaulters is discussed in this paper. Dataset from the UCI machine learning repository comprising of 25 attributes and 30000 instances have been employed to analyze the performance of algorithms. Moreover, the effect of feature selection has also been identified with respect to each classification algorithm. It has been concluded from the experimental results that both Correlation Feature Subset and Information Gain feature selection methods yield the most useful features for prediction and the accuracy of Random Forest Ensemble method is highest in predicting credit card defaulters. General Terms Data Mining, Credit card defaulters dataset.