Analyzing Sequential Patterns in Customer Supermarket Transactions

This paper seeks to examine sequential pattern mining techniques for discovering recurring patterns of purchasing in a retail supermarket environment. The objective is to understand the interconnectivity between different products and their co-purchasing trends through analysis of transaction data using the Apriori algorithm and other pattern discovery models. Our approach is based on data preprocessing from raw transactional data. Then, we apply the Apriori algorithm for the generation of frequent itemsets and compare the execution time for different algorithms like Apriori and FP-Growth. Results show FP growth take less time to execute. We further apply clustering methods and use ARIMA time-series forecasting on customer segments to predict the trend of sales in the near future. The results will help retailers formulate an improved marketing strategy, correct placement of products, and interaction with customers.