COMPARATIVE STUDY OF CLUSTERING ALGORITHMS AND AN IMPLEMENTATION OF THE APRIORI ALGORITHM IN A RETAIL COMPANY
Muhammad Rakhmat Setiawan, 14522440
MetadataShow full item record
At the retail company, several problems often arise from the sale of the products. The retail company found difficulty to perform the correct promotion strategy for improving their product sales. The information about the customers' segmentation will be beneficial to determine the right business strategy for the retail company. In the study, a detailed discussion on each of three algorithms, i.e., Hierarchical, K Means and Fuzzy C Means Clustering Algorithms and a comparative study between them already investigated. Other than that, the implementation of Association Rules Market Basket Analysis (ARMBA) by using Apriori Algorithm is applied to uncover the associations between transactions. After conducting 100 simulations, the researcher found that Fuzzy C Means is the best clustering algorithm regarding the measure of goodness with the BSS/TSS ratio of 59,34% and average processing time equal to 63,23 seconds. Otherwise, K Means Algorithm has BSS/TSS ratio of 56,7% and average processing time corresponding to 93.83 seconds. There are two customer clusters in which Customer Cluster A and Customer Cluster B with different emphasize to be considered by the business owner. The characteristics of each cluster with these two approaches are the Customers Cluster A tend to buy adult products, and the Customers Cluster B tend to buy baby products. The researcher found that the most promising customers are in Customer Cluster A regarding the comparison of each variable on clustering algorithms result. ARMBA Apriori Algorithm with minimum confidence 80% and support 0,1% has generated 23 rules. Based on these results, therefore, the researcher proposed to the business owner to focus on customer cluster A and provide ARMBA results using Apriori Algorithm so that later is expected to be the basis to formulate business strategy by the business owner.
- Industrial Engineering