Machine Learning Strategy for Improved Prediction of Micronutrient Concentrations in Soils of Taif Rose Farms Based on EDXRF Spectra

This study attempts to utilize newly developed machine learning techniques in order develop a general prediction algorithm for agricultural soils Saudi Arabia, specifically the Taif region. Energy dispersive X-ray fluorescence (EDXRF) measurements were used national predictive models that predict concentrations of 14 micronutrients rose farms, providing high-quality data comparable conventional methods. Machine algorithms this included simple linear model, multivariate regression (MLR); and two nonlinear models, random forest (RF) adaptive splines (MARS). Our proposes (ML) strategy predicting fertility parameters more accurately using 10 farms (Rosa damascena) Taif, Arabia as case study. Results demonstrated MARS provides higher performance when number explanatory variables is small, while RF superior large. On other hand, MLR recommended moderate method variables. The showed can be overwhelm drawbacks EDXRF device, such high detection limits an element cannot directly measured.