Supermarket Data Analysis Based on Non-Negative Matrix Factorization

The data generated from supermarket shopping in people's daily life. The information in the data reflects the customer's behavioral preference. Mining this type of information can help us better understand customer behaviors, and then supermarket has more marketing strategies to make more profits and win the competition in the face of fierce competition, so the practical significance is very strong. But according to the existing research on the supermarket data, we can find that there is no unified method to analyze the supermarket data. In view of this problem, this paper establishes a new data analysis framework, which takes non-negative matrix factorization (NMF) algorithm and supermarket data analysis as the research object. It attempts to use the NMF method to analyze the customer purchasing behavior. Since the interpretation result of NMF method is easy to interpret, the method has good interpretability. The study uses an algorithm that combines traditional non-negative matrix factorization and Monte Carlo initial method, which attempts to apply this algorithm to larger datasets. In the future, it can be integrated with more data analysis methods to get more analytical information.