Oil palm is a vital force in driving the energy business. In 2020, Thailand had 9,954.27 sq.km. (around 6,220,799 Rai) of oil palm plantations, ranking third in the world after Indonesia and Malaysia. Ranong has the highest oil palm crop yield per Rai in Thailand. Notwithstanding, it is challenging to classify land use accurately and keep it up to date by using only labor, due to the need for a number of laborers and high labor costs. Moreover, land use/land cover cannot use spectral information classification alone. Nevertheless, machine learning is a popular data estimation technique that enables a system to learn from sample data; however, there are few studies on its use for data fusion techniques in order to classify land use/land cover, especially concerning oil palm. Therefore, we aim to apply machine learning and data fusion to classify land use/land cover, especially for oil palm. After a multicollinearity test of spectral information and ancillary variables, Surface Reflectance (SR) of Blue, Near Infrared, SWIR-1, NDWI, NDVI and LST were selected with a threshold of correlation coefficients. A stepwise stack of six inputs was created. The first stack included only Surface Reflectance (SR) of Blue, Near Infrared and SWIR-1. NDWI, NDVI and LST were added later. ID4 (Surface Reflectance (SR) of Blue, Near Infrared, SWIR-1, NDWI, NDVI and LST) in the random forest model resulted in OA being 0.9341 and KC being 0.9239, which was the highest among 12 models. ID4 in the random forest model provided the classification results for oil palm very close to the factual number per the figure of 2.90 sq.km (around 1,814 Rai) from the Department of Land.