Empirical Evaluation of Machine Learning Techniques for Car Price Prediction

Recently there has been a sudden increase in the demand for second-hand cars as consumers have been unable to afford brand-new cars due to a variety of factors, including high pricing, limited availability, financial inability, etc. The second-hand automobile industry, however, is still in its development phase and is mostly controlled by the informal sector. When buying a second-hand automobile, creates the possibility of fraud. In order to predict the price of a used automobile without favoring the consumer or the merchandiser, a high-accuracy model is needed. To obtain high accuracy, a variety of regression algorithms such as support vector regression (SVM), linear regression, random forest regression, decision tree and polynomial regression are utilized. R-squared was calculated to evaluate how well each regression performed. Of all the regressions used, Random Forest’s R-squared was the highest. Here are some potential areas of novelty that were further explored to improve the accuracy of the Random Forest model to 0.9234: