An integrated methodology with data mining techniques for retail industry

Over the past decade, organizations has struggled with huge amount of data since advanced data collection and storage tools have unprecedentedly been improving along with technology. Particularly, retail industry is the leading sector in terms of generating a vast number of transaction data. Hence, Data Mining is a good solution for extracting useful knowledge from the collected data. This study aims to cluster stores of a retail company by using data mining methodology. Various data mining techniques have been applied to group stores according to similarities of personnel cost. Initially, K-means which is a prominent clustering algorithm was applied for clustering, however it could not cope with particular problems due to nature of data and the problem. For this reason, an integrated methodology including Random Forest and K-medoids is proposed to overcome the faced problems in K-means application. Statistical software programs, Weka (3.6) and R (3.1.2), were used to run these algorithms. The study was carried out in LC Waikiki which is a large retail company serving in textile industry. Although the company has a great number of stores throughout Turkey and several countries, a pilot study including 85 stores in İstanbul is considered for the applications.