Data-Driven Prediction of Vehicle Residual Value in the UAE Automotive Industry

In this master’s thesis, machine learning methods are applied to predict the residual value of a vehicle, focusing on the problem of estimating the vehicle’s price after sale. The research addresses some of the most important drivers of depreciation which include market segmentation, country, region-specific details, transmission type, and some numerical values obtained through factor analysis. This is very important to the automotive industry, leasing businesses, and financial institutions. The data was collected from the open-source online marketplace Dubizzle and contains 25 variables and 62,922 cases. The initial data posed notable challenges as there were non-standardized feature values, incomplete data fields, and inconsistent categorizations leading to assortment which demanded tremendous preprocessing efforts. The study adheres to the CRISP-DM (Cross-Industry Standard Process for Data Mining) model for data cleansing, structuring, and analysing. Vehicle residual value was predicted using Generalized Linear Regression, Least Squares Support Vector Machine (LSVM), and Neural Networks. The Neural Network model achieved the best results with a correlation of 0.988 and a relative error of 0.024, outperforming LSVM (correlation: 0.953, error: 0.092) and Regression (correlation: 0.936, error: 0.124) models, which were slightly lower in performance. The results validate that market trends, vehicle characteristics, and several hidden quantitative elements greatly influence retaining vehicle value. These findings address the emerging role of machine learning in the fields of automotive valuation and price prediction. Additional parameters like real-time market data, macroeconomic trends, and consumer sentiment indicators could be added to improve the prediction models in future research. The research builds on automotive analytics by constructing a model for vehicle depreciation estimation based on available data, which aids in pricing and evaluating financial risk.