Machine-Learning-Based Assessment and Prediction of Digital Transformation Performance for MSMEs in Indonesia

Assessing digital capabilities, including literacy, software, and infrastructure, is crucial in the micro, small, and medium enterprises (MSMEs) digital transformation. This study uses machine learning to assess and predict digital transformation performance for MSMEs in Indonesia. The research stages consist of cluster analysis using the Gaussian Mixture Model, latent variables analysis using Factor Analysis of Mixed Data, assessment analysis, prediction model development using Random Forest, and analysis of digital transformation performance using Shapley Additive exPlanations (SHAP). Findings reveal that MSMEs’ digital transformation relies on key variables: Quick Response Code Indonesian Standard (QRIS) adoption, e-commerce engagement, e-payment use, and online community participation. Machine learning categorizes MSMEs into four readiness levels: intention/beginner, adopter, performer, and transformer. The experimental results show a high classification accuracy with an average accuracy of 97%. The study outcomes can assist policymakers in evaluating and crafting strategic development plans for the digital transformation of MSMEs in Indonesia.