Enhancing Code Smell Detection Performance in Python Programming Language : A Comparative Study

Code smell is one of the problems in programming which indicates that a problem has occurred, where there is something less than ideal in the code even though the code can run well. This research conducted a comparative study of the performance of Decision Tree, Random Forest, the use of the AdaBoost, CatBoost, XGBoost ensemble, and the use of SMOTEENN preprocessing to improve code smell detection in the Python programming language. Overall, the Decision Tree Pruning model hybrid with Adaboost and SMOTEEN produces the highest accuracy of 98.69% and MCC of 97.40%. Meanwhile, on the Long Method dataset, the XGBoost model with the SMOTEENN application produces the highest accuracy of 99.69% and MCC of 99.38%.