AUTOMATED PRICING PREDICTIONS FOR PRE-OWNED VEHICLES USING RANDOM FOREST

Abstract

The pricing of new vehicles in the automotive industry is determined by manufacturers, augmented by additional costs imposed by the government in the form of taxes. Consequently, customers purchasing a new car can be assured that their investment is justified. However, due to escalating prices of new vehicles and the financial constraints faced by many consumers, the market for used cars is experiencing significant global growth. This underscores the pressing necessity for an effective used car price prediction system that can accurately assess the vehicle's value based on various factors, including mileage, year of manufacture, fuel consumption, transmission type, road tax, fuel type, and engine size. We have developed a highly effective model designed to serve the needs of sellers, buyers, and manufacturers within the used car market. Upon completion, this model will deliver relatively precise price predictions based on the information provided by users. The Random Forest algorithm was employed in this research to maximize accuracy, enabling the prediction of an actual vehicle price rather than merely a price range. To evaluate the performance of each regression model, the R-squared metric was calculated.