Sinta Devi Rahmawati, Adinda Bintang Oktavia, Fadina Salwa Aulia Putri, Diana Laily Fithri

Penerapan Algoritma Apriori Untuk Menemukan Pola Asosiasi Pada Data Penjualan Retail Fashion

Introduction

Penerapan algoritma apriori untuk menemukan pola asosiasi pada data penjualan retail fashion. Temukan pola asosiasi pembelian di retail fashion menggunakan algoritma Apriori pada 3.400 transaksi. Optimalkan promosi, rekomendasi produk, dan tata letak toko strategis.

Abstract

Penelitian ini menerapkan algoritma Apriori dalam analisis data transaksi penjualan di bidang retail fashion guna menemukan keterkaitan produk yang kerap dibeli dalam satu waktu. Dataset yang dianalisis berisi 3.400 transaksi pelanggan dari platform Kaggle, dan diolah menggunakan RapidMiner dengan parameter minimum support 0,1 serta confidence 0,6. Metode yang digunakan meliputi preprocessing data, normalisasi, transformasi one-hot encoding, dan pengujian dengan operator W-Apriori. Hasilnya ditemukan pola signifikan, seperti pembelian backpack dan loafers berasosiasi kuat dengan raincoat (confidence 74%). Algoritma Apriori terbukti efisien dalam mengenali pola kebiasaan pembelian konsumen, dan dapat digunakan untuk kegiatan promosi, penyusunan rekomendasi produk, dan penataan layout toko. Aturan asosiasi yang diperoleh mencerminkan pola perilaku konsumen saat berbelanja, salah satunya menunjukkan adanya hubungan erat antara beberapa produk. Informasi tersebut dapat dimanfaatkan sebagai dasar pengambilan keputusan strategis yang berorientasi data, termasuk dalam penataan produk yang lebih efisien, pengembangan fitur rekomendasi, serta perancangan promosi yang relevan dan tepat sasaran.

Review

This paper presents a focused application of the Apriori algorithm to uncover association patterns within fashion retail sales data, addressing a relevant challenge for businesses seeking to optimize their strategies. Utilizing a dataset of 3,400 customer transactions from Kaggle and processed through RapidMiner, the research aims to identify products frequently purchased together. A key finding highlights a strong association between the purchase of backpacks and loafers with raincoats, exhibiting a confidence level of 74%. The study successfully demonstrates how association rule mining can extract actionable insights from transactional data, providing a practical foundation for data-driven decision-making in retail. The methodology employed, which includes data preprocessing, normalization, one-hot encoding, and testing with the W-Apriori operator, appears sound and appropriate for the stated objective. The choice of the Apriori algorithm is well-justified for discovering 'if-then' relationships in transaction databases, and its application in the fashion retail context offers tangible benefits. The paper effectively articulates the practical implications of its findings, such as informing promotional activities, enhancing product recommendation systems, and optimizing store layouts. These contributions underscore the study's potential to add significant value to retail management by translating raw sales data into strategic business intelligence. While the study provides a clear demonstration of Apriori's utility, a few areas could further strengthen its contribution. A more detailed description of the dataset, beyond just the transaction count, such as the number of unique items or categories, would provide better context. Justification for the selected minimum support (0.1) and confidence (0.6) thresholds, possibly accompanied by a sensitivity analysis to illustrate their impact on the generated rules, would enhance methodological robustness. Furthermore, while one specific association is highlighted, discussing the overall number of rules found and perhaps presenting a broader range of interesting patterns, or even the limitations encountered, would enrich the discussion. Future research could explore the integration of temporal dynamics or comparative analyses with other association rule mining algorithms.

Full Text

Comments

You need to be logged in to post a comment.