FP-Growth Implementation for Market Basket Analysis in Building Materials Store in Surabaya

Abstract

Building materials store generally manage large volumes of transaction data with unique characteristics such as high product variety and uneven purchase distribution making these data often not fully used for business decisions effectively. This study analyses customer purchasing patterns and proposes product bundling recommendations specifically for building material sector using Market Basket Analysis with FP-Growth algorithm. The research uses 46,533 transactions records collected from a building materials store in Surabaya between July and September 2025. FP-Growth was selected for its efficiency in handling sparse data without candidate generation compared to traditional algorithm like Apriori. Due to the uneven distribution of product purchases, a minimum support of 0.5% and a minimum confidence of 0.3 were selected experimentally to obtain relevant and interpretable results. The strength of the relationships between products was evaluated using support, confidence, and lift values, identifying strong association in complementary items like nails and cement mixtures. To validate the results, interviews with the store owner and questionnaires for customers were conducted. CSAT was used to assess the accuracy of the purchasing patterns, while conversion rate measured customer interest in the proposed bundles. Results indicate a high validation score with CSAT reaching 75% and a potential conversion rate of up to 81% for specific bundles. These findings demonstrate that the proposed data-driven approach is highly suitable for practical implementation in optimizing retail marketing strategies.Building materials store generally manage large volumes of transaction data with unique characteristics such as high product variety and uneven purchase distribution making these data often not fully used for business decisions effectively. This study analyses customer purchasing patterns and proposes product bundling recommendations specifically for building material sector using Market Basket Analysis with FP-Growth algorithm. The research uses 46,533 transactions records collected from a building materials store in Surabaya between July and September 2025. FP-Growth was selected for its efficiency in handling sparse data without candidate generation compared to traditional algorithm like Apriori. Due to the uneven distribution of product purchases, a minimum support of 0.5% and a minimum confidence of 0.3 were selected experimentally to obtain relevant and interpretable results. The strength of the relationships between products was evaluated using support, confidence, and lift values, identifying strong association in complementary items like nails and cement mixtures. To validate the results, interviews with the store owner and questionnaires for customers were conducted. CSAT was used to assess the accuracy of the purchasing patterns, while conversion rate measured customer interest in the proposed bundles. Results indicate a high validation score with CSAT reaching 75% and a potential conversion rate of up to 81% for specific bundles. These findings demonstrate that the proposed data-driven approach is highly suitable for practical implementation in optimizing retail marketing strategies.