Variable: fNRec Ranger result Call: ranger(formulaString.lst[[j]], data = dfs, importance = "impurity", write.forest = TRUE, mtry = t.mrfX$bestTune$mtry, num.trees = 500) Type: Regression Number of trees: 500 Sample size: 5 Number of independent variables: 377 Mtry: 18 Target node size: 5 Variable importance mode: impurity OOB prediction error: 843.4912 R squared: -0.2553873 OOB RMSE: 29.043 Variable importance: [,1] AAIavg_GYGA 0 af_agg_30cm_AWCpF23__M_1km 0 af_agg_30cm_PWP__M_1km 0 af_agg_30cm_TAWCpF23__M_1km 0 af_agg_30cm_TAWCpF23mm__M_1km 0 af_agg_30cm_TETAs__M_1km 0 af_agg_ERZD_TAWCpF23mm__M_1km 0 af_BDRICM_T__M_1km 0 af_ERZD__M_1km 0 Al_M_agg30cm_AF_1km 0 ASSDAC3 0 B02CHE3 0 B04CHE3 0 B07CHE3 0 B13CHE3 0 eXtreme Gradient Boosting 5 samples 377 predictors No pre-processing Resampling: Cross-Validated (3 fold, repeated 1 times) Summary of sample sizes: 3, 4, 3 Resampling results across tuning parameters: eta max_depth nrounds RMSE Rsquared 0.3 2 50 16.42154 1 0.3 2 100 16.42132 1 0.3 2 150 16.42132 1 0.3 3 50 16.42154 1 0.3 3 100 16.42132 1 0.3 3 150 16.42132 1 0.3 4 50 16.42154 1 0.3 4 100 16.42132 1 0.3 4 150 16.42132 1 0.3 5 50 16.42154 1 0.3 5 100 16.42132 1 0.3 5 150 16.42132 1 0.3 6 50 16.42154 1 0.3 6 100 16.42132 1 0.3 6 150 16.42132 1 0.3 7 50 16.42154 1 0.3 7 100 16.42132 1 0.3 7 150 16.42132 1 0.3 8 50 16.42154 1 0.3 8 100 16.42132 1 0.3 8 150 16.42132 1 0.4 2 50 16.42129 1 0.4 2 100 16.42129 1 0.4 2 150 16.42129 1 0.4 3 50 16.42129 1 0.4 3 100 16.42129 1 0.4 3 150 16.42129 1 0.4 4 50 16.42129 1 0.4 4 100 16.42129 1 0.4 4 150 16.42129 1 0.4 5 50 16.42129 1 0.4 5 100 16.42129 1 0.4 5 150 16.42129 1 0.4 6 50 16.42129 1 0.4 6 100 16.42129 1 0.4 6 150 16.42129 1 0.4 7 50 16.42129 1 0.4 7 100 16.42129 1 0.4 7 150 16.42129 1 0.4 8 50 16.42129 1 0.4 8 100 16.42129 1 0.4 8 150 16.42129 1 0.5 2 50 16.42132 1 0.5 2 100 16.42132 1 0.5 2 150 16.42132 1 0.5 3 50 16.42132 1 0.5 3 100 16.42132 1 0.5 3 150 16.42132 1 0.5 4 50 16.42132 1 0.5 4 100 16.42132 1 0.5 4 150 16.42132 1 0.5 5 50 16.42132 1 0.5 5 100 16.42132 1 0.5 5 150 16.42132 1 0.5 6 50 16.42132 1 0.5 6 100 16.42132 1 0.5 6 150 16.42132 1 0.5 7 50 16.42132 1 0.5 7 100 16.42132 1 0.5 7 150 16.42132 1 0.5 8 50 16.42132 1 0.5 8 100 16.42132 1 0.5 8 150 16.42132 1 Tuning parameter 'gamma' was held constant at a value of 0 Tuning parameter 'colsample_bytree' was held constant at a value of 0.8 Tuning parameter 'min_child_weight' was held constant at a value of 1 RMSE was used to select the optimal model using the smallest value. The final values used for the model were nrounds = 100, max_depth = 2, eta = 0.4, gamma = 0, colsample_bytree = 0.8 and min_child_weight = 1. RMSE: 16.421 R2: 1 XGBoost variable importance: Feature Gain Cover Frequency 1: AAIavg_GYGA 9.909450e-01 0.78571429 0.78571429 2: af_agg_30cm_PWP__M_1km 9.052973e-03 0.17857143 0.17857143 3: af_agg_30cm_TAWCpF23mm__M_1km 1.991695e-06 0.03571429 0.03571429 4: NA NA NA NA 5: NA NA NA NA 6: NA NA NA NA 7: NA NA NA NA 8: NA NA NA NA 9: NA NA NA NA 10: NA NA NA NA 11: NA NA NA NA 12: NA NA NA NA 13: NA NA NA NA 14: NA NA NA NA 15: NA NA NA NA Ensemble validation RMSE: 19.217 R2: 0.318 -------------------------------------- Variable: fPRec Ranger result Call: ranger(formulaString.lst[[j]], data = dfs, importance = "impurity", write.forest = TRUE, mtry = t.mrfX$bestTune$mtry, num.trees = 500) Type: Regression Number of trees: 500 Sample size: 7 Number of independent variables: 377 Mtry: 16 Target node size: 5 Variable importance mode: impurity OOB prediction error: 39.94965 R squared: 0.4352421 OOB RMSE: 6.321 Variable importance: [,1] NCluster_11_AF_1km 6.681056 L10USG5 6.078063 PHIHOXagg0_30 5.972478 M43BSALT 5.848447 ECN_M_agg30cm_AF_1km 5.782310 AfSIS_WRB_RefGc10 5.724592 ENTENV3 5.670176 Zn_M_agg30cm_AF_1km 5.496004 Rice_intermed 5.445234 M13RB3A08 5.370037 Rice_irrigated_high_baseline 5.312803 B13CHE3 5.135677 ENAX_M_agg30cm_AF_1km 5.111473 Temperature 5.085312 EXMOD5avg 5.038442 eXtreme Gradient Boosting 7 samples 377 predictors No pre-processing Resampling: Cross-Validated (3 fold, repeated 1 times) Summary of sample sizes: 5, 4, 5 Resampling results across tuning parameters: eta max_depth nrounds RMSE Rsquared 0.3 2 50 7.669297 1 0.3 2 100 7.668320 1 0.3 2 150 7.668320 1 0.3 3 50 8.514510 1 0.3 3 100 8.513531 1 0.3 3 150 8.513531 1 0.3 4 50 8.500779 1 0.3 4 100 8.499802 1 0.3 4 150 8.499802 1 0.3 5 50 8.514646 1 0.3 5 100 8.513667 1 0.3 5 150 8.513667 1 0.3 6 50 8.488622 1 0.3 6 100 8.487644 1 0.3 6 150 8.487644 1 0.3 7 50 8.488725 1 0.3 7 100 8.487747 1 0.3 7 150 8.487747 1 0.3 8 50 8.487363 1 0.3 8 100 8.486385 1 0.3 8 150 8.486385 1 0.4 2 50 8.526874 1 0.4 2 100 8.526860 1 0.4 2 150 8.526860 1 0.4 3 50 8.577670 1 0.4 3 100 8.577658 1 0.4 3 150 8.577658 1 0.4 4 50 8.588767 1 0.4 4 100 8.588754 1 0.4 4 150 8.588754 1 0.4 5 50 8.486159 1 0.4 5 100 8.486146 1 0.4 5 150 8.486146 1 0.4 6 50 8.543562 1 0.4 6 100 8.543549 1 0.4 6 150 8.543549 1 0.4 7 50 8.588950 1 0.4 7 100 8.588937 1 0.4 7 150 8.588937 1 0.4 8 50 8.503494 1 0.4 8 100 8.503481 1 0.4 8 150 8.503481 1 0.5 2 50 8.579123 1 0.5 2 100 8.579123 1 0.5 2 150 8.579123 1 0.5 3 50 8.561191 1 0.5 3 100 8.561191 1 0.5 3 150 8.561191 1 0.5 4 50 8.487623 1 0.5 4 100 8.487623 1 0.5 4 150 8.487623 1 0.5 5 50 8.571418 1 0.5 5 100 8.571418 1 0.5 5 150 8.571418 1 0.5 6 50 8.494236 1 0.5 6 100 8.494236 1 0.5 6 150 8.494236 1 0.5 7 50 8.588969 1 0.5 7 100 8.588969 1 0.5 7 150 8.588969 1 0.5 8 50 8.563738 1 0.5 8 100 8.563738 1 0.5 8 150 8.563738 1 Tuning parameter 'gamma' was held constant at a value of 0 Tuning parameter 'colsample_bytree' was held constant at a value of 0.8 Tuning parameter 'min_child_weight' was held constant at a value of 1 RMSE was used to select the optimal model using the smallest value. The final values used for the model were nrounds = 100, max_depth = 2, eta = 0.3, gamma = 0, colsample_bytree = 0.8 and min_child_weight = 1. RMSE: 7.668 R2: 1 XGBoost variable importance: Feature Gain Cover Frequency 1: af_agg_30cm_PWP__M_1km 5.474375e-01 0.53551913 0.46666667 2: ASSDAC3 2.236001e-01 0.03825137 0.03333333 3: af_BDRICM_T__M_1km 1.509867e-01 0.16393443 0.15000000 4: B13CHE3 6.885438e-02 0.05737705 0.05000000 5: C01GLC5 7.422568e-03 0.01912568 0.01666667 6: AAIavg_GYGA 1.511878e-03 0.15300546 0.23333333 7: af_agg_30cm_TAWCpF23mm__M_1km 1.868484e-04 0.02185792 0.03333333 8: af_agg_30cm_TAWCpF23__M_1km 9.693515e-09 0.01092896 0.01666667 9: NA NA NA NA 10: NA NA NA NA 11: NA NA NA NA 12: NA NA NA NA 13: NA NA NA NA 14: NA NA NA NA 15: NA NA NA NA Ensemble validation RMSE: 6.833 R2: 0.305 -------------------------------------- Variable: fKRec Ranger result Call: ranger(formulaString.lst[[j]], data = dfs, importance = "impurity", write.forest = TRUE, mtry = t.mrfX$bestTune$mtry, num.trees = 500) Type: Regression Number of trees: 500 Sample size: 6 Number of independent variables: 377 Mtry: 18 Target node size: 5 Variable importance mode: impurity OOB prediction error: 26.74573 R squared: 0.6145162 OOB RMSE: 5.172 Variable importance: [,1] M13RB3A04 8.182305 Rice_intermed 7.238613 C01GLC5 5.252749 PHIHOXagg0_30 5.246155 VDPMRG5 4.714012 NIRL14 4.698626 NCluster_4_AF_1km 4.402626 REDL14 4.360590 M13RB1A04 4.261162 Temperature 4.243461 EVEENV3 4.181063 AfSIS_WRB_RefGc10 4.105696 NCluster_19_AF_1km 4.014402 af_BDRICM_T__M_1km 4.008058 L10USG5 3.987804 eXtreme Gradient Boosting 6 samples 377 predictors No pre-processing Resampling: Cross-Validated (3 fold, repeated 1 times) Summary of sample sizes: 4, 4, 4 Resampling results across tuning parameters: eta max_depth nrounds RMSE Rsquared 0.3 2 50 4.471256 1 0.3 2 100 4.467616 1 0.3 2 150 4.467616 1 0.3 3 50 4.471256 1 0.3 3 100 4.467616 1 0.3 3 150 4.467616 1 0.3 4 50 4.471256 1 0.3 4 100 4.467616 1 0.3 4 150 4.467616 1 0.3 5 50 4.471256 1 0.3 5 100 4.467616 1 0.3 5 150 4.467616 1 0.3 6 50 4.471256 1 0.3 6 100 4.467616 1 0.3 6 150 4.467616 1 0.3 7 50 4.471256 1 0.3 7 100 4.467616 1 0.3 7 150 4.467616 1 0.3 8 50 4.471256 1 0.3 8 100 4.467616 1 0.3 8 150 4.467616 1 0.4 2 50 4.467764 1 0.4 2 100 4.467654 1 0.4 2 150 4.467654 1 0.4 3 50 4.467764 1 0.4 3 100 4.467654 1 0.4 3 150 4.467654 1 0.4 4 50 4.467764 1 0.4 4 100 4.467654 1 0.4 4 150 4.467654 1 0.4 5 50 4.467764 1 0.4 5 100 4.467654 1 0.4 5 150 4.467654 1 0.4 6 50 4.467764 1 0.4 6 100 4.467654 1 0.4 6 150 4.467654 1 0.4 7 50 4.467764 1 0.4 7 100 4.467654 1 0.4 7 150 4.467654 1 0.4 8 50 4.467764 1 0.4 8 100 4.467654 1 0.4 8 150 4.467654 1 0.5 2 50 4.467692 1 0.5 2 100 4.467692 1 0.5 2 150 4.467692 1 0.5 3 50 4.467692 1 0.5 3 100 4.467692 1 0.5 3 150 4.467692 1 0.5 4 50 4.467692 1 0.5 4 100 4.467692 1 0.5 4 150 4.467692 1 0.5 5 50 4.467692 1 0.5 5 100 4.467692 1 0.5 5 150 4.467692 1 0.5 6 50 4.467692 1 0.5 6 100 4.467692 1 0.5 6 150 4.467692 1 0.5 7 50 4.467692 1 0.5 7 100 4.467692 1 0.5 7 150 4.467692 1 0.5 8 50 4.467692 1 0.5 8 100 4.467692 1 0.5 8 150 4.467692 1 Tuning parameter 'gamma' was held constant at a value of 0 Tuning parameter 'colsample_bytree' was held constant at a value of 0.8 Tuning parameter 'min_child_weight' was held constant at a value of 1 RMSE was used to select the optimal model using the smallest value. The final values used for the model were nrounds = 100, max_depth = 2, eta = 0.3, gamma = 0, colsample_bytree = 0.8 and min_child_weight = 1. RMSE: 4.468 R2: 1 XGBoost variable importance: Feature Gain Cover Frequency 1: af_agg_30cm_PWP__M_1km 9.644093e-01 0.74358974 0.74358974 2: ASSDAC3 2.020619e-02 0.05128205 0.05128205 3: af_BDRICM_T__M_1km 1.533543e-02 0.12820513 0.12820513 4: B13CHE3 4.819582e-05 0.02564103 0.02564103 5: af_agg_30cm_TAWCpF23mm__M_1km 8.731479e-07 0.02564103 0.02564103 6: AAIavg_GYGA 4.129466e-08 0.02564103 0.02564103 7: NA NA NA NA 8: NA NA NA NA 9: NA NA NA NA 10: NA NA NA NA 11: NA NA NA NA 12: NA NA NA NA 13: NA NA NA NA 14: NA NA NA NA 15: NA NA NA NA Ensemble validation RMSE: 5.61 R2: 0.565 --------------------------------------