Implementation of the Apriori Algorithm for Purchase Pattern Analysis

Canda Steak & Café menghadapi tantangan dalam memanfaatkan data transaksi untuk mengidentifikasi pola pembelian konsumen yang mendukung strategi bundling dan cross-selling. Penelitian ini mengimplementasikan algoritma Apriori dari awal menggunakan Python untuk merancang platform web interaktif berbasis Streamlit, dan mengevaluasi aturan asosiasi dengan lima metrik komprehensif (support, confidence, lift, leverage, conviction). Metodologi Rapid Application Development (RAD) diterapkan dengan tahapan perencanaan kebutuhan, desain sistem, pengembangan, implementasi, dan evaluasi. Validasi terhadap pustaka MLxtend menunjukkan akurasi 100% pada frequent itemsets dan association rules dengan waktu eksekusi 0,3326 detik untuk 1.000 transaksi. Hasil analisis mengidentifikasi pola asosiasi kuat antara menu utama dan snack pendamping dengan confidence hingga 80% dan lift 1,60, menghasilkan rekomendasi strategi bundling dengan estimasi ROI 15-20%.

Canda Steak & Café faces challenges in utilizing transaction data to identify consumer purchasing patterns that support bundling and cross-selling strategies. This research implements the Apriori algorithm from scratch using Python to design an interactive web platform based on Streamlit, and evaluates association rules with five comprehensive metrics (support, confidence, lift, leverage, conviction). The Rapid Application Development (RAD) methodology is applied with stages of requirements planning, system design, development, implementation, and evaluation. Validation against the MLxtend library showed 100% accuracy on frequent itemsets and association rules with an execution time of 0.3326 seconds for 1,000 transactions. The analysis results identified strong association patterns between main menus and side snacks with confidence up to 80% and lift of 1.60, resulting in bundling strategy recommendations with an estimated ROI of 15-20%.