Extraction of Sales Relevance from POS Data Based on Non negative Matrix Factorization with Regularization  
Author Sota Hamaya


Co-Author(s) Yuka Minamino; Taku Moriyama; Mio Hosoe; Masashi Kuwano


Abstract Point-of-sales data are a class of big data used for improving the operational efficiency and promoting the product sales of companies and organizations. Analyzing the sales data of a wide variety of products and understanding the sales trends and characteristics of individual products consume unrealistically large calculation resources. The present study proposes non negative matrix factorization with regularization for classifying products into similar groups and identifying the sales trends of each group. When applied to the POS data of an antenna shop in Japan, non negative matrix factorization with regularization extracted five characteristic product groups and their time-series patterns.


Keywords Non Negative Matrix Factorization, Regularization, POS Data
    Article #:  RQD27-92

Proceedings of 27th ISSAT International Conference on Reliability & Quality in Design
Virtual Event

August 4-6, 2022