Implementing Market Basket Analysis Using Eclat And Apriori Algorithm On Grocery Product Marketing Strategy

Market basket analysis is the disclosure of the associations between the items and based on those associations it helps to analyze consumer behavior. The transactional data have been recurrently used as an object of research as means of revelation of new information for example in winter people wear comfortable warm outfits while in summer people wear comfortable half sleeves. Similarly, consumer purchase behavior also changes with the change in time. We scrutinize the problem of discovering the association between items. We will study the relationship between association rules and time. We have majorly focused on methods of finding the effective minimum threshold value, which plays a significant role in giving precise results. Its value highly influences the number of items in the final item set. Various methods have been used, but the idea of dynamic threshold value performs best as it changes its value after every step of the process. The two major algorithms used in frequent rule mining are Apriori and Eclat algorithms. Both of the algorithms have different approaches to giving results and have different efficiency in different sizes of datasets. The characteristics of the dataset highly affect the performance of algorithms. The small dataset best suits the algorithm while the large dataset mostly uses Apriori. The Eclat algorithm is the most efficient in its depth-first search technique which lets it scan the dataset only once whereas Apriori scans the dataset in every step which increases the time complexity and space complexity for storing the result of each step. Both algorithms have their pros and cons, which we highlight in this study.