Analysis of the random forest and grid search algorithms in early detection of diabetes mellitus disease

Andi Andi (Institut Bisnis Informasi Teknologi dan Bisnis, Indonesia)
Thamrin Thamrin (Institut Bisnis Informasi Teknologi dan Bisnis, Indonesia)
Agus Susanto (Institut Bisnis Informasi Teknologi dan Bisnis, Indonesia)
Elyzabeth Wijaya (Institut Bisnis Informasi Teknologi dan Bisnis, Indonesia)
Deva Djohan (Institut Bisnis Informasi Teknologi dan Bisnis, Indonesia)

Jurnal Mantik
Vol. 7 No. 2 (2023): Agustus: Manajemen, Teknologi Informatika dan Komunikasi (Mantik)

This research focuses on implementing the Random Forest and Grid Search algorithms for the early detection of diabetes mellitus, aiming to modernize and enhance medical practices using technology. The proposed model achieved an accuracy of 77.06%, a precision of 71.43%, a recall of 47.30%, and a misclassification error of 22.94%. Comparative analysis with other data mining algorithms, including Decision Tree, Random Forest without Grid Search, and Cat Boost, demonstrated that the Random Forest with Grid Search algorithm outperformed the others. By utilizing Grid Search, the accuracy of the Random Forest algorithm increased by 2.03%. These findings indicate the potential effectiveness of machine learning in early diabetes detection. While the research offers promising results, there are limitations in terms of the dataset size and the number of detection variables used. Future studies should explore larger datasets and alternative algorithms to further enhance accuracy and aid in the early detection of diabetes mellitus.