Research On Product Sales Forecast Considering Test Review Data

Product testing is a product sales method often used by companies.Its purpose is to open up product sales channels,increase product sales,or to facilitate the design of advertisements.Through product testing,on the one hand,the market potential can be judged by the sales of the product so as to make actual production arrangements.On the other hand,companies can learn about the shortcomings of the product from the reviews of consumers during the testing period.In addition,the reviews may also contain valuable product improvement suggestions.The company can improve the product based on these data.This research predicts the product sales based on the sales and review data during the testing period of an e-commerce website.First of all,based on the existing research,the review factors and product factors that affect the product sales are obtained,and the sales forecast model is constructed in combination with the sales during the testing period.Then,feature extraction is performed on product reviews,and the BERT model is used to judge the emotion tendency of the reviews and identify product attributes.Finally,a sales forecast model is constructed based on the extracted review features,sales during the testing period and product features,and the effect of the forecast model is evaluated.The results of this research show that review emotion and review response have a significant positive impact on product sales,while the number of reviews,the length of reviews,and review differences have a negative impact on product sales.In addition,the emotion tendency of product features contained in product reviews does have a significant impact on product sales.Among the prediction models constructed by the research,the multiple linear regression model has the worst effect,the prediction effect of random forest and BP neural network is similar,and the prediction effect of XGBoost model is the best.In addition,compared with the regression model that does not consider product feature emotion,the adjusted R~2 of the regression model considering product feature emotion is larger,and the model has a better fit for the prediction of product sales.This research also conducted a robustness test,which shows that the method used in this study is robust.