Menganalisa Pola Pembelian Pelanggan Dengan Metode Data Mining: Customer Segmentation dan Market Basket Analysis Menggunakan Algoritme K-Means dan Eclat

Analyzing Customer Purchasing Patterns Using Data Mining Methods: Customer Segmentation and Market Basket Analysis Using K-Means and Eclat Algorithm

Data mining merupakan suatu proses mencari pola atau informasi yang menarik pada suatu data menggunakan teknik atau metode tertentu. Dengan menerapkan data mining toko swalayan dapat mengolah data transaksinya menjadi informasi yang berharga guna meningkatkan bisnisnya. Dua metode data mining yang sering digunakan dalam mengelola data transaksi yang besar yaitu customer segmentation dan market basket analysis.  Metode customer segmentation dapat mengelompokkan pelanggan yang memiliki kesamaan perilaku menjadi beberapa cluster. Sedangkan metode market basket analysis dapat mengolah data transaksi tiap cluster menjadi beberapa aturan yang kuat, dimana aturan ini dapat digunakan untuk mengidentifikasi peluang penjualan silang (cross-selling) dan penjualan tambahan (up-selling).

Dalam penelitian ini, data yang digunakan diperoleh dari CV. Aloha Ponorogo pada tahun 2022 sebanyak 997 data pelanggan dan 22.359 data transaksi. Data tersebut diproses menggunakan metode customer segmentation dengan algoritme k-means melalui bahasa pemrograman R menghasilkan 5 cluster berdasarkan kesamaan sifat dan karakteristik pelanggannya dengan jumlah kelompok cluster pertama sebanyak 217 pelanggan, kelompok cluster kedua sebanyak 209 pelanggan, kelompok cluster ketiga sebanyak 295 pelanggan, kelompok cluster keempat sebanyak 233 pelanggan, dan kelompok cluster kelima sebanyak 41 pelanggan. Dan selanjunya untuk tiap-tiap cluster dianalisa menggunakan metode market basket analysis dengan algoritme eclat dengan nilai minimal support sebesar 3.5% dan nilai minimal confidence sebesar 10% diperoleh hasil bahwa tiap cluster memiliki aturan asosiasi yang berbeda beda, pada cluster pertama mengahasilkan 8 aturan asosiasi, cluster kedua menghasilkan 4 aturan asosiasi, cluster ketiga menghasilkan 2 aturan sosiasi, cluster keempat menghasilkan 10 aturan asosiasi, dan cluster kelima menghasilkan 18 aturan sosiasi.

Outup dari hasil penelitian kedua metode tersebut dapat direkomendasikan beberapa hasil penataan produk dengan tujuan untuk menambah penjualan (up-selling) dan penjualan silang (cross-selling).

Kata Kunci – Data Mining, Customer Segmentation, Market Basket Anaysis, K-Means, Eclat, R.

Data mining is a process of looking for interesting patterns or information in data using certain techniques or methods. By implementing data mining, supermarkets can process their transaction data into valuable information to improve their business. Two data mining methods that are often used in managing large transaction data are customer segmentation and market basket analysis. The customer segmentation method can group customers who have similar behavior into several clusters. Meanwhile, the market basket analysis method can process transaction data for each cluster into several strong rules, where these rules can be used to identify cross-selling and additional sales (up-selling) opportunities.

In this research, the data used was obtained from CV. Aloha Ponorogo in 2022 will have 997 customer data and 22,359 transaction data. The data is processed using the customer segmentation method with the k-means algorithm through the R programming language to produce 5 clusters based on the similarities in the nature and characteristics of customers with the first cluster group totaling 217 customers, the second cluster group having 209 customers, the third cluster group having 295 customers, the cluster group the fourth had 233 customers, and the fifth cluster group had 41 customers. And then, for each cluster, it was analyzed using the market basket analysis method with the Eclat algorithm with a minimum support value of 3.5% and a minimum confidence value of 10%. The results showed that each cluster had different association rules, the first cluster produced 8 association rules, the second cluster produces 4 association rules, the third cluster produces 2 association rules, the fourth cluster produces 10 association rules, and the fifth cluster produces 18 association rules.

The output from the research results of these two methods can be recommended for several product structuring results with the aim of increasing sales (up-selling) and cross-selling.

Keyword – Data Mining, Customer Segmentation, Market Basket Anaysis, K-Means, Eclat, R.