New Energy Vehicles Sales Prediction Model Combining the Online Reviews Sentiment Analysis: A Case Study of Chinese New Energy Vehicles Market

Abstract

AbstractResearches on the sales of new energy vehicles (NEVs) can provide theoretical support and practical instructions for the government and the automobile industry. Traditional sales prediction methods often solely includes historical sales data or excludes the impact of public sentiment. In this paper Random Forest Regression (RFR) model is trained on the historical influencing factors and monthly sales using supervised learning method. Then, SARIMA model is used to predict the influencing factors. Lastly, these SARIMA predicted influencing factors are passed into the trained RFR model to acquire the predicted monthly sales and compare with true monthly sales. For further improvement, text sentiment variables are introduced into the regression model. The results indicate that the RFR model shows better performance in predicting monthly sales when taking online comments of the month before last month as text sentiment variables.KeywordsNew energy vehicles (NEVs)Sales predictionRandom Forest Regression (RFR)SARIMAText sentiment variables