Research on sales forecasting in e-commerce industry for imbalanced classification data

Abstract

The background of this study is that with the advent of the big data era, e-commerce sales forecasting has become a key factor in improving the market competitiveness and economic benefits of enterprises. To solve this problem, we used machine learning technology to build a comprehensive sales forecasting system. By processing massive sales data, including data cleaning, label encoding, outlier processing and other steps, we established a complete data set. In terms of model selection, we tried multiple regression models, such as RandomForestRegressor [1], ExtraTreesRegressor, etc., and evaluated their performance through cross-validation. In order to solve the problem of data imbalance, a combination of oversampling technology (RandomOverSampler)[2] and normalization processing is used. Finally, we selected ExtraTreesRegressor as the best model and evaluated it on the training set. The research results show that the accuracy and reliability of sales forecasts can be improved by comprehensively processing sales data and selecting appropriate machine learning models. The contribution of this study in the field of e-commerce sales forecasting is to provide a comprehensive and practical solution, which provides important decision-making support for enterprises in market competition. Combining machine learning technology and data processing methods, we provide e-commerce companies with an effective sales forecasting strategy that is expected to have a positive impact in improving market competitiveness, reducing risk costs, and accelerating revenue growth [3]. Future research directions can be carried out in deeply exploring the characteristics of sales data, optimizing model parameter adjustment, and combining professional knowledge in more fields. Introducing more emerging machine learning algorithms and technologies to adapt to the changing market demands in the e-commerce field is expected to further improve the performance and adaptability of the sales forecasting system.