A study of sales forecasting in multinational retail companies: a feature extraction-machine learning-classification based forecasting framework

This paper proposes a framework for accurately predicting future sales volume in retail enterprises using advanced data mining techniques and algorithms. The framework includes feature extraction, machine learning, and classification prediction based on traditional machine learning. The paper introduces the general design of the framework and the basic theories of the models used, including correlation analysis, time-series models, and machine learning. The data used for analysis is statistically described, and relevant features are extracted. The framework is trained and tested using the data, and the results are compared with those of other models. The paper concludes that the proposed framework outperforms XGBoost alone and provides a new idea for sales data prediction in the retail industry.