<![CDATA[This study analyzes customer purchasing patterns at Hatfina Hijab and translates the resulting association rules into actionable Business Intelligence (BI) recommendations. The dataset consists of 174 sales transactions recorded from February to April 2024. After data cleaning, product items were transformed into a binary transaction matrix and processed using association rule mining in RapidMiner. The baseline rule set was obtained using a minimum support of 0.30 and a minimum confidence of 0.50. The revised analysis complements the original confidence values with support and lift to distinguish rules that are frequent from those that represent genuinely positive product associations. Nineteen baseline rules were reported, of which sixteen have lift values greater than 1.00 and therefore indicate positive associations. The strongest confidence was found in the rule Hampers Paket Hemat 2 Hijab and Hampers Paket Hemat 1 Hijab -> Souvenir Sajadah (confidence = 1.000; lift = 1.475). Parameter sensitivity analysis shows that stricter support-confidence thresholds reduce the retained rule set from 19 rules at 0.30/0.50 to 14 rules at 0.35/0.60, 13 rules at 0.40/0.70, 3 rules at 0.45/0.80, and 1 rule at 0.50/0.90. The findings are interpreted into recommendations for product bundling, cross-selling, promotional design, and coordinated inventory planning. The study demonstrates how association rule mining can function as an analytical component of BI and Decision Support Systems (DSS), while emphasizing that association patterns indicate co-occurrence rather than causality.]]>
Copyrights © 2026
