Application of K-Means Clustering for Mapping and Analyzing Small and Medium Industry Potential Based on Production, Investment, and Labor Characteristics M. Azwan, Muhammad Syahputra Novelan

Abstract

Small and Medium Industries (SMIs) play an important role in supporting regional economic
growth, creating employment opportunities, and strengthening local economic development.
However, differences in investment value, workforce size, and production performance create
challenges in identifying and mapping the potential of each industrial unit. Therefore, a
systematic analytical approach is required to group industries based on their characteristics
and provide useful information for regional development planning. This study aims to apply
the K-Means Clustering algorithm to map and analyze the potential of Small and Medium
Industries based on production, investment, and labor characteristics. The research uses
secondary data from the SMI database in Tanjungbalai City, North Sumatra, Indonesia,
covering the 2021–2026 period. The initial dataset consists of 709 industrial records, with 708
complete records selected for the clustering process. Three numerical attributes are used as
the main clustering variables, namely workforce size, investment value, and production value.
The research stages include data collection, preprocessing, feature selection, data
transformation, standardization, K-Means implementation, and cluster evaluation. The
clustering results produce three groups: 502 records (70.90%) classified as low potential, 195
records (27.54%) as medium potential, and 11 records (1.55%) as high potential. Model
evaluation generates a Davies-Bouldin Index (DBI) of 0.814 and a Silhouette Coefficient of
0.436. These findings indicate that K-Means Clustering can identify differences in industrial
characteristics and provide structured information regarding the distribution of SMI potential.
The results are expected to support government institutions in determining development
priorities, improving investment capacity, strengthening workforce utilization, and
formulating more targeted industrial development strategies based on data-driven decisionmaking.