Classification of Credit Card Default Clients Using LS-SVM Ensemble

摘要

Finding knowledge from a database and turning it into useful information is a big challenge. The use of machine learning helps analyze data and contribute to delivering results that can be acted upon by the company. SVM is one of machine learning method that has better performance than other machine learning method but sensitive to parameter setting and training sample. the performance accuracy of the SVM method can be improved using the LS-SVM and Ensemble method. This research proposes the LS-SVM ensemble to identify the prospective credit cards client that will default. The Least Square SVM ensemble method has the highest percentage with a difference of 1.7% from SVM and 0.6% from Least Square SVM.