Analisis Loyalitas Pelanggan Berbasis Model Recency, Frequency, dan Monetary (RFM) dan Decision Tree pada PT. Solo

Abstract

Perkembangan bisnis alat tulis kantor dan sekolah saat ini banyak yang menjanjikan, maka banyak bermunculan pemasok baru dalam bisnis Alat Tulis Kantor dan Sekolah (ATKS). PT Solo yang bergerak di bidang bisnis ATKS harus memiliki strategi dalam setiap persaingan usaha, khususnya dalam meraih loyalitas pelanggan. Loyalitas pelanggan sering dipengaruhi oleh faktor jumlah aktivitas transaksi, nilai nominal transaksi, waktu transaksi di perusahaan, dan atribut outlet. Penelitian ini mengusulkan model Recency, Frequency, dan Monetary (RFM) yang dikombinasikan dengan Decision Tree. Model RFM digunakan untuk proses klasterisasi data pelanggan berdasarkan jumlah transaksi, nilai nominal transaksi, waktu transaksi, dan atribut outlet. Sedangkan Decision Tree dapat menggambarkan tingkat loyalitas pelanggan. Data transaksi dalam penelitian ini dilakukan sepanjang 1 Januari hingga 31 Desember 2018 terhadap 1.203 pelanggan dan 18.087 transaki melalui faktur pembelian. Hasil penelitian ini menunjukan bahwa state-of-the-art pada model RFM dan Decision Tree yang diusulkan lebih unggul dibandingkan hanya dengan menggunakan model RFM saja. Cluster ke-1 memiliki 860 pelanggan menghasilkan loyalitas pelanggan sedang (biru), cluster ke-2 memiliki 69 pelanggan menghasilkan loyalitas pelanggan yang tinggi (hijau), dan cluster ke-3 memiliki 274 pelanggan menghasilkan loyalitas pelanggan yang rendah (merah). Model klasterisasi RFM dan klasifikasi Decision Tree telah menghasilkan atribut outlet yang berpengaruh terhadap nilai akurasi sebesar 67,54%. Abstract The development of office and school stationery business at this time, many promising, so many new suppliers have sprung up in the office and school stationery business. PT Solo, which has the office and school stationery business, must have a strategy in every business competition, especially in achieving customer loyalty. Customer loyalty is often influenced by factors in the number of transaction activities, transaction nominal value, transaction time at the company, and outlet attributes. This research proposes a Recency, Frequency, and Monetary (RFM) model combined with a Decision Tree. RFM model is used to process customer data clustering based on number of transactions, transaction nominal value, transaction time, and outlet attributes. Whereas Decision Tree can describe the level of customer loyalty. Transaction data in this study were conducted from 1 January to 31 December 2018 to the 1,203 customers and 18,087 transactions through purchase invoices. The results of this study indicate that the state-of-the-art in the proposed RFM and Decision Tree models is outperform compared to only using the RFM model. Cluster 1 has 860 customers resulting in moderate customer loyalty (blue), Cluster 2 has 69 customers resulting in high customer loyalty (green), and Cluster 3 has 274 customers resulting in lower customer loyalty (red). RFM clustering model and Decision Tree classification have produced outlet attributes that affect the accuracy value of 67.54%.