In this work, various statistical learning methods are utilized to examine the performance changes of the bankruptcy prediction model over time. This study is conducted on companies listed on the Korea Exchange(KRX) from January 1, 2000 to January 31, 2016, and the bankrupted enterprises are companies that request for delisting or workout.We consider accounting variables indicating the potential growth of the company and the Index of All Industrial Productions (IAIP) as macroeconomic variable. Additional market variables are chosen based on Market Capitalization of each company; Variability, Skewness, and Kurtosis. Let the years during the January 2000?January 2016 period be a seqence of times. Then, we compose some pairs of train and test data. First three years are designated as training data, and the consequent next one year is set as testing data. For each three-one pair trial, we predict the bankruptcy of the enterprise and conduct a series of tests on all trials. For example, we take train data for 2000-2002 and test data for 2003 for the analysis. This process is repeatedly carried out. Using Logistic Regression, Partial Least Squares (PLS), and Random Forest, the final model is selected by comparing the recall, a measure of classification performance evaluation. Logistic Regression and Random Forest assign weights on bankruptcy companies due to imbalance of the bankruptcy data. Hence, to overcome the fact that the Partial Least Squares method is impossible to utilize weights itself, we accept the Synthetic Minority Oversampling Techniques (SMOTE) to consider the data properties. Consequently, the result reveals that accounting variables and IAIP indirectly affects on the bankruptcy, and the random forest model, non-statistical learning methods, demonstrates more significant results than other methods in a series of tests.