<![CDATA[East Java Province is one of the major agricultural production centers in Indonesia with diverse biophysical characteristics. Variations in topography, climate, soil properties, and vegetation conditions contribute to differences in agricultural land suitability across the region. This study aimed to analyze agricultural land suitability in East Java Province using a Random Forest algorithm integrated with Geographic Information Systems (GIS) and multisource geospatial data. The variables employed included elevation, slope, aspect, rainfall, temperature, soil pH, soil organic carbon (SOC), Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Normalized Difference Water Index (NDWI). Data were obtained from SRTM, CHIRPS, TerraClimate, SoilGrids, Landsat 8, and ESA WorldCover datasets. The Random Forest model was developed using 80% of the samples for training and 20% for testing. Model evaluation showed an overall accuracy of 80.12%, a Cohen’s Kappa coefficient of 0.4212, and an ROC-AUC value of 0.8596, indicating good classification performance. Feature importance analysis revealed that slope was the most influential variable (18.43%), followed by NDWI (14.69%), elevation (12.24%), and NDVI (11.63%). Spatial analysis indicated that moderately suitable land dominated East Java, covering approximately 3.28 million ha (48.71%), followed by land of low suitability, covering 2.00 million ha (29.75%), and highly suitable land, covering 1.45 million ha (21.54%). The results demonstrate that integrating Random Forest with GIS provides an effective approach for assessing agricultural land suitability and can support sustainable agricultural planning and spatial decision-making in East Java Province]]>
