Second-hand Cars Recommender System Model using the SMOTE and the Random Forest Technique

Abstract

This research aims to develop a model for a second-hand cars recommender system towards the use of the SMOTE approach together with the Random Forest technique. The research team has modified the imbalance of the data by employing the SMOTE method to increase the amount of data to 400% and applying the Random Forest technique to create a decision tree for second-hand car recommendations. After evaluating the model’s effectiveness by using the 10-fold validation method, the findings revealed that the second-hand recommender system model applying the SMOTE approach and the Random Forest technique provided an accuracy of 98.84%, a precision of 98.89%, a recall of 98.80% and an F1 score of 98.80%; the overall scores of the model’s effectiveness were higher than of the model using only the Random Forest technique. This indicates that the model can be practically used for recommending second-hand cars.