STUDI KOMPARATIF ALGORITMA C4.5 DAN RANDOM FOREST PADA DIGITALISASI UMKM KABUPATEN TEGAL

Abstrak

Digital transformation has become an essential necessity for Micro, Small, and Medium Enterprises (UMKM) to enhance their competitiveness in the era of Industry 4.0. However, in Tegal Regency, the level of digitalization adoption among MSMEs remains varied and tends to be low, thus requiring further investigation. This study aims to compare the performance of the C4.5 and Random Forest algorithms in classifying the level of digitalization of MSMEs in Tegal Regency. This research employs the CRISP-DM methodology, which includes business understanding, data understanding, data preparation, modeling, evaluation, and implementation. Primary data were collected through questionnaires distributed to 100 MSME respondents and processed using RapidMiner. The results indicate that the Random Forest algorithm demonstrates superior performance, achieving an average F1-score of 89.63%, accuracy of 91.43%, while the C4.5 algorithm records an average F1-score of 86.24%, accuracy of 90%. The highest F1-score for both algorithms is observed in the low digitalization category at 95%, which is consistent with the data distribution showing that the majority of MSMEs (59%) fall within this category. This study systematically integrates the CRISP-DM approach from business understanding to model implementation, resulting in a structured data analysis workflow that can be replicated by local governments or future researchers. Another novelty of this study lies in the finding that although Random Forest exhibits better classification performance than C4.5, the majority of MSMEs remain at a low level of digitalization. These results provide practical contributions as a basis for formulating more targeted and sustainable MSME digitalization policies at the local