Usage of Ensemble Regression Technique for Product Price Prediction
T. Bhaskar, S. Arumai Shiney, S. Babitha Rani, K. Maheswari, Samrat Ray, V. Mohanavel
Abstract
In the fast-moving world, people like brokers and third-party vendors easily deceive the people who try to buy used products like electronics and automobiles. As many buyers will not be educated about the necessary parameters which are to be considered before buying a product. This makes it easier for them to sell products that are not in a proper working condition. To resolve this issue, this study aims the development of three different regression models which are capable of predicting the selling price of used cars based on the input parameters like actual price, kilometers traveled, etc. Three different regression techniques are used in the construction of the regression models. The techniques are linear regression, lasso regression, and ensemble regression. The ensemble regression is the combination of both linear regression and lasso regression. The models are then trained using the preprocessed dataset which will be obtained from Kaggle. The preprocessing techniques include null value elimination and data encoding. The trained models are then tested to find the best model. The accuracy and the error values of the regression models were compared. The results of the comparison provide the best regression model which can be employed in the prediction of the price of the used cars.