Penerapan Data Mining dan Machine Learning dalam Menentukan Strategi Bisnis UMKM Berdasarkan Tren Penjualan dan Perilaku Konsumen

M. Dico TriyadI (Universitas Pembangunan Panca Budi)
Muhammad Syahputra Novelan (Universitas Pembangunan Panca Budi)
Rian Farta Wijaya (Universitas Pembangunan Panca Budi)

Instal : Jurnal Komputer
Vol 18 No 5 (2026): Instal : Jurnal Komputer

This study aims to utilize Big Data and Machine Learning technology to assist the strategic decision-making process for MSMEs at the Deli Serdang Regency Cooperatives and SMEs Office. The methods applied include ARIMA and SARIMA to forecast sales trends, K-Nearest Neighbor (KNN) for consumer behavior classification and segmentation, and Rough Set for feature selection. The data used includes sales transaction data and consumer behavior for the period 2020–2023. The study shows that the Rough Set method can reduce data dimensionality by 50% without reducing model accuracy. The KNN model shows excellent performance with a high evaluation value (accuracy reaching 0.99), and can group consumers into loyal, potential, and non-loyal categories. On the other hand, the ARIMA and SARIMA models show unsatisfactory performance with high error rates, making them less appropriate for fluctuating data. Overall, the combination of Data Mining and Machine Learning has proven efficient in generating strategic information that can support MSMEs in formulating data-driven business strategies.