A Comparative Study on Used Car Price Prediction Model

Abstract

The number of used car prices in the market keeps increasing due to the launch of a new model by the car manufacturer. The sales price is determined by the car's specifications and present state. The objective of this study is to make a comparison of the machine learning algorithm that can be implemented for used car price prediction. Previous studies on used car price prediction commonly perform the comparison of the machine learning model, meanwhile, the study on stock price prediction utilizes hyperparameter tuning. It shows that hyperparameter tuning can increase the performance of the machine learning model. The expected outcome from the study is the
best machine learning model will be used for used car price prediction. The machine learning model will be trained by using Azure Machine Learning Studio. Therefore, the study compares four different machine learning models, including linear regression, neural network regression, boosted decision tree regression, and decision forest regression. As a result, boosted decision tree regression is indicated as the most effective model, exhibiting high R-squared values and superior performance compared to the other machine learning algorithms. This study also will perform a comparative study on the prediction model with hyperparameter tuning to get the most feasible and accurate model for the prediction model. There are two different hyperparameter tuning being compared such as the entire grid and random sweep and the study shows that random sweep provides the best R-squared values at 0.874548. The best machine learning model will be deployed and integrated with the web application that is developed by using ASP.NET. This study will be beneficial to the community in providing valuable insight into the factors that influence used car prices. These insights can be utilized by industry professionals and market analysts to make informed decisions and develop effective strategies.