Research on Supermarket Marketing Data Analysis Based on Business Intelligence

Abstract

In recent years, with the rapid development of the new retail industry, consumers have more comparison and choice when purchasing goods, which leads to increasingly fierce competition in the supermarket industry and continuous compression of profit space. If you want to improve the competitiveness of supermarkets, you can conduct business intelligence analysis and sales forecast on a large number of data generated by supermarket operation and management, thus providing an important basis for supermarket operation and management strategy adjustment. This paper uses the marketing data of a global supermarket for four years as the data base, analyzes the current business situation from different angles, uses python to conduct data preprocessing, analysis and visualization, and explores the sales strategy to improve sales through sales analysis, commodity analysis and user analysis. It uses the data to find new growth points, and obtains methods to further improve the supermarket sales. Finally, the integrated learning algorithms XGBoost, lightGBM and RandomForest in machine learning are used to build a prediction model and extract four different types of feature set data. The average score values predicted by the three models for ‘Sales’ are different. Among the four types of feature set data, the Average Score value obtained from RandomForest is higher than XGBoost and lightGBM models, and the Average Score value obtained from the “sub_cate_all” feature set data is higher than the value obtained from the other three feature set data, which is 81.25%, indicating that RandomForest has the best prediction effect among the three models.