Research On Laptop Price Predictive Model Based on Linear Regression, Random Forest and Xgboost

Abstract

In today's rapidly evolving technological environment, the portability and versatility of laptops have led to a significant growth in their user base as well, making them essential tools for individuals and businesses alike. Particularly since the COVID-19 pandemic, the rate of Work From Home (WFH) partially or overall WFH has exceeded 30%. Global laptop shipments have also far surpassed desktop PCs so far in 2017. Accurately predicting laptop prices is beneficial for retailers to devise competitive pricing strategies and for consumers to effectively budget and select the most suitable laptops. In this study, we used a dataset of 1320 samples to investigate the significance of features in a laptop price prediction model using linear regression, random forest, and XGBoost methods. We incorporated 13 features in the modeling process, including laptop brand, type, screen size, RAM, GPU, operating system, and weight. Three mathematical models were established to predict the price of laptop. By comparing the regression model metrics RMSE and R2 under linear regression, random forest and XGBoost models, the RMSE under the XGBoost model is 294.11 with an R2 of 0.85. It is evident that the XGBoost model exhibits the smallest RMSE and the highest R2 value closest to 1. This suggests that the XGBoost model provides the highest accuracy and best fit for the predictive model.