Segmentation and Prediction of the Development of Small and Medium Industries (SMEs) in North Sumatra Using K-Means and Random Forest Algorithms Based on OSS RBA Data

Abstract

Small and Medium Industries (SMEs) have a strategic role in increasing regional economic growth, creating jobs, and encouraging the competitiveness of the national industrial sector. Along with the implementation of the Online Single Submission Risk-Based Approach (OSS RBA) system, business licensing data is available that can be used to support data-based decision-making. However, the use of OSS RBA data in identifying the characteristics and development of SMEs is still limited. This study aims to segment the development of SMEs in North Sumatra Province using the K-Means algorithm and build a prediction model using the Random Forest algorithm. The research dataset comes from OSS RBA data for the 2022-2025 period which has gone through the stages of data cleaning, feature engineering, data aggregation, and normalization. The number of clusters was determined using the Elbow Method, Silhouette Score, and Davies-Bouldin Index, then the segmentation process was carried out using the K-Means algorithm. The results of the segmentation are then used as the classification target in the Random Forest model. The results showed that the optimal number of clusters used was five clusters (K=5). The Random Forest model is able to predict segmentation results with excellent performance based on the results of the model evaluation. The Feature Importance analysis shows that the variables of Year, Growth of SMEs, Number of SMEs, and several characteristics of the industrial sector are the most influential factors in the formation of clusters. The results of this research are expected to be the basis for the formulation of policies for fostering and developing SMEs based on the characteristics of the region and industrial sector in North Sumatra Province.