Residual Value Prediction

Abstract

Car leasing is an important business sector. The residual value is the value of the car at the end of the lease. The residual value determines the monthly payment in a car leasing contract. Predicting the residual value of a car accurately is important for the car leasing company. In this paper, we investigate using machine learning techniques to carry out residual value prediction. We developed seven residual value prediction models using Lasso Regression, Decision Tree, Random Forest, Light GBM, XGBoost, CatBoost and Neural Network. We evaluated and compared the performance of these models using the data collected from a financial service company in New Zealand. Our experience show that the model based on CatBoost achieves the best accuracy in terms of mean absolute error and mean absolute percentage error. Compared with the method currently used by the financial service company, the CatBoost-based model reduces the prediction error by 50%.