Automobile Price Prediction using Regression Models

Abstract

Regression models are used to forecast automobile prices with the purpose of supporting a new entrant into the industry. Because it takes substantial knowledge in the subject, predicting the price of an automobile has become a popular research topic. The present work develops an automobile prediction system of price, with the help of supervised regression. For this, data cleaning has been done by converting null values of some features into non-null values to enhance the performance of regression model. Five regression models i.e. linear regression, random forest, decision tree, elastic net, and SVR are used for comparing the price prediction of automobile. Dataset used for this purpose is obtained from Kaggle having price of 205 different automobiles with 26 features. The automobile price prediction has been analyzed with three performance parameters i.e. R2 score, Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). Out of the five regression models, random forest regression model has performed best with values 0.93 as R2 Score, 1390.9 as MAE and 2139.7 as RMSE.