Research on Demand Forecasting Method of Multi-user Group Based on Big Data

Abstract

In order to accurately meet the purchasing needs of consumers, this paper proposes a multi-user demand forecasting model based on big data that organically combines sentiment classification and user portraits. The study takes the online reviews of smart watches on an e-commerce website as the data source, the product attributes that users pay attention to are obtained through word frequency analysis and LDA model, and the NLPIR sentiment analysis tool is used to analyze their sentiment tendency to construct a user demand evaluation system; then count the word frequency of perceptual words, classify them with kJ analysis method, so as to mine the perceptual needs of users, and use the Censydiam model to explore the user's purchasing motivation and perform crowd clustering, and finally build user portraits; then count the scores of each user group on the demand evaluation indicators, extract the product design objectives and distinguish their importance according to the functional positioning and application strategy of the indicator type, and establish the demand forecasting model of multi-user groups. The research results show that through data mining and perceptual engineering analysis, we can get the improvement trend of products in the future, make them better meet the needs of users, and provide effective guidance for product design.