Indonesian Retail Market Analysis Using Frequent Pattern Data Mining

Abstract

IT plays an enormous role in creating conditions for faster socioeconomic development of a nation. Computerized data analysis methods can provide vital information for many businesses to make informed decisions in time and stay competitive. Frequent pattern (FP) mining is a data mining approach for discovering useful market data automatically from seemingly abundant data that are difficult to analyze using conventional statistical approaches. FP approaches have been widely accepted in developed countries for market data analysis and consumer behavior analysis, but seldom evaluated or applied in developing countries, such as Indonesia, due to limited access to data in the regions. However, such data would be tremendously useful for investors and financial institutions who are planning to invest in those developing countries. For this reason, we have collected large itemset transaction data from Indonesia for over 2 years and devised FP mining approaches to explore the quality of the data and extract useful data for foreign and local investors. For the first time, this paper reports the FP analysis results, new analysis approaches for this emerging market using novel datasets. In particular, we show that the average number of items that are bought together frequently in a retail shop in Indonesia is three. This kind of information is useful for formulating various marketing strategies, but difficult to obtain using conventional statistical approaches.