Mining Association Rules in Seasonal Transaction Data

Nowadays global retail business faces new challenges due to changes in consumer buying preferences, likewise in Indonesia that consequence increasing level of competition among retailers. Moreover, Indonesia's government announced major reforms in foreign investment in 2016 which attracted international retailers. Therefore, local retailers have to plan a strategy to expand the current market through enhancement of customer satisfaction related to their purchasing activities. This research aims to analyze market basket data to help local retailers understand consumer purchase behavior by finding association pattern. Nevertheless, the retail industry is highly seasonal. Hence, there are three types of season in retail industry which were proposed as classification in this research, i.e. peak season, normal season, and slack season. Based on the result, each season has similarity in generated pattern among other seasons. However, there are also 6 unique patterns found from a certain season. By knowing the discovered association patterns in each season, the company may determine product offering or promotion strategy for different season.