Fig 3. R_c²,RMSE_c and MAE_c distributions of the machine learning regression models based on PVIs (MLR-PVIs models) in differrent spectral dataset.

R_c2,RMSE_c and MAEc represente determination coefficient, root mean square error and mean absolute error of calibration respecttively. R, 1/R, logR, R^1/2,R′ and R′′ represent original spectra, reciprocal spectra, logarithmic spectra, square root spectra, first-drivative and second-derivative spectra respectively. AdaBoost, ET and RF represent Adaptive Boosting models, Extra Trees models and Random forest models respectively.