Soil Organic Carbon Prediction Using Digital Color Sensor in Peru

Montero E.; Vásquez A.; Alayo L.; Gutiérrez P.; Mestanza C. · Communications in Computer and Information Science (2023) · DOI: 10.1007/978-3-031-35445-8_16

Resumen

Soil organic carbon traditional analysis methods have a high economic, environmental and time-consuming cost. General efforts are focused on developing new low-cost and environmentally friendly alternatives. The aim was to evaluate the prediction of soil organic carbon using machine learning for local conditions of Peru. A total of 351 dry topsoil samples were analyzed to obtain soil organic carbon and color measurements. The models used were multiple linear regression, generalized additive models, regression tree cubist, random forest, K-nearest neighbors and neural networks. The results show that the simplest models were competitive with complex models, suggesting to incorporate other predictor variables to improve the models performance.

Instituciones: Universidad Nacional Agraria La Molina