@article{oai:tokyo-metro-u.repo.nii.ac.jp:00009434,
 author = {Srisomkiew, Sasirin and Kawahigashi, Masayuki and Limtong, Pitayakon},
 journal = {Geographical reports of Tokyo Metropolitan University},
 month = {},
 note = {Using machine learning (ML) algorithms to digital soil mapping (DSM) allows the
elucidation of relationships between soil properties and environmental variables enabling the precise prediction of soil nutrient levels. The accuracy of the predicted values using the random forest (RF) algorithm, which is the most popular ML algorithm for digital soil mapping, and multiple linear regression (MLR) were compared to create digital maps of soil chemical properties in the Thung
Kula Ronghai (TKR) region, Thailand. The spectral indices including moisture stress index (MSI), normalized difference water index (NDWI), saturation index (SI), brightness index (BI), and coloration index (CI) obtained from remote sensing (RS) data were found to be more effective for predicting the various soil properties than the topographic indices derived from the DEM in the plain area. The MLR and RF models successfully predicted soil chemical properties with good predictive
accuracy. The results indicated that the RF model has a slightly higher accuracy than the MLR model.
However, the MLR model is superior in interpreting the relationship with the model equations.},
 pages = {1--11},
 title = {Comparison between random forest and multiple linear regression to create digital maps of soil chemical properties in the Thung Kula Ronghai region, Thailand},
 volume = {57},
 year = {2022}
}