Type-2 diabetes (DM2) is still growing into a global problem. Decreased insulin sensitivity is the main mechanism causing DM2. PPAR-gamma has been shown to have a direct relationship with insulin sensitivity. This study aimed to develop a machine-learning model to evaluate phytochemical compounds’ activity as a PPAR-gamma agonist. The dataset used for modeling was procured from the ChEMBL database. A total of 3668 substances were retained after null and duplicate values were removed. Canonical SMILES were collected and described to generate an extended connectivity fingerprint, which was then used to predict the pEC50 value. The machine learning-generated model was then used to predict 2846 phytochemical compounds from the KEGG database. This study uses the Light gradient-based machine learning method. The performance of the model was evaluated using Root Mean Square Error (RMSE) and R-square (R2) of 0.63 and 0.73, respectively. Afterward, the model was used to predict phytochemical compounds. Five compounds with the lowest EC50 were obtained: Pedilstatin (60.05 nM), Gnididin (64.29 nM), Paclitaxel (98.93 nM), Vincristine (113.9 nM), and Camellidin II (129.72 nM). Further research regarding the direct effect of PPAR-gamma needs to be tested on these five compounds. Machine learning models will save time and resources for drug discovery or repurposing previously used drugs.