Year: 2025
Type: conference-paper
Abstract: In view of the low accuracy and difficulty of new product sales prediction,a new product sales prediction model based on the XGBoost algorithm is proposed, which integrates the sentiment analysis results of customer reviews to improve the accuracy of new product sales prediction. The large language model is used to perform text processing and analysis on customer reviews to obtain product attributes and sentiment characteristics related to the product; the sentiment characteristics are associated with the historical sales of new products to construct a sales prediction model based on the XGBoost algorithm . In order to evaluate the performance of the model, four new cigarette products of a certain brand in 2023 were selected for verification. The results show that the model proposed in this paper is superior to the traditional model in both MAE and RMSE evaluation indicators. Compared with the basic XGBoost model, MAE decreased by 26.88% and RMSE decreased by 18.54%, indicating that sentiment characteristics help improve prediction accuracy.
Authors: Bin Lu, Chao Deng, Xuexia Liang, Xipeng Liu, Liang Zeng
Institution: China Tobacco Guangxi Industrial (China)
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W4414288773