Product Reviews Analysis of E-commerce Platform Based on Logistic-ARMA Model

With the vigorous development of 5G Internet technology, people’s demand for purchasing goods through ecommerce platform is daily increasing. E-commerce platforms now mostly launch commodity evaluation function, which will undoubtedly affect other consumers’ choice of goods, but it can also provide a reference for platforms to adjust sales strategies at any time. At present, many large transnational e-commerce platforms are committed to scientific commodity evaluation trend prediction, so as to timely understand the popularity of commodities based on the existing data, so as to maximize their own interests. Amazon’s platform provides consumers with product ratings and text-based reviews. We collected the recent evaluation data of the three commodities on the platform, and put forward reasonable analysis and prediction. Firstly, the text-based evaluation and commodity rating are processed quantitatively to form a two-dimensional evaluation vector, and the part that is inconsistent with the evaluation emotion is eliminated to realize data cleaning. We use the Logistic Regression Equation Model to complete the function fitting based on time variables for the three commodities, and achieve the statistical analysis of the existing data. In addition, we draw on the principle of filtering out noise outside the band in signal processing, and make innovative use of Trend Moving Average Model (ARMA) to smooth the function to eliminate the impact of various random events on commodity evaluation as much as possible, so as to realize the prediction of the positive Trend of future evaluation. In this paper, an ecommerce platform based on customer’s evaluation of products to achieve a high sensitivity, constructive project for future product sales forecast.