Random Forest Result #144 — Analysis Report
Contents
- random forest (#144) completed
1. Random Forest
2026-09-21 02:14:14 · random_forest
The model explains 53.9% of the variance in the target under cross-validation.
| task_type | regression |
|---|---|
| r2_mean | 0.538756 |
| r2_std | 0.214189 |
| mse_mean | 14.986814 |
| mse_std | 12.519635 |
permutation importance
| name | mean | std |
|---|---|---|
| PLOT | 0.011272 | 0.008572 |
| ENV_SOILOC | 0.001235 | 0.007825 |
| ENV_SOILTC | 0.002577 | 0.011296 |
| ENV_SOILTN | 0.102671 | 0.070685 |
| ENV_PH | -0.002838 | 0.002858 |
| ENV_SOILCLAY | 0.01445 | 0.009704 |
| ENV_SOILSILT | 0.006406 | 0.023445 |
| ENV_SOILSAND | 0.010353 | 0.00754 |
| ENV_SOILMOIS | 0.021319 | 0.018914 |
| ENV_SOILP | 0.000114 | 0.000991 |
shap importance
| name | mean_abs |
|---|---|
| PLOT | 0.1131 |
| ENV_SOILOC | 0.183659 |
| ENV_SOILTC | 0.126229 |
| ENV_SOILTN | 0.327007 |
| ENV_PH | 0.109797 |
| ENV_SOILCLAY | 0.061339 |
| ENV_SOILSILT | 0.167248 |
| ENV_SOILSAND | 0.065622 |
| ENV_SOILMOIS | 0.205682 |
| ENV_SOILP | 0.03121 |
split metrics
| r2 | mse |
|---|---|
| 0.308397 | 5.958141 |
| 0.483595 | 6.311223 |
| 0.824277 | 32.691079 |
importance bar
permutation bar
shap summary
pred vs obs
cv boxplot
Method & Provenance
- Method
- Random Forest (Regression)
- Engine
- Python · scikit-learn 1.8.0
- Input dataset
- gallery — catg_example_03_soil_nitrogen_availability.csv
- Started
- 2026-09-21 02:14:14.068246
- Permanent link
- https://cat-g-omxpert.wishartlab.com/results/by-token/413ee895cd9742ed95bd1502567c309a
| Parameter | Value |
|---|---|
| task_type | regression |
| model_type | regression |
| target_column | ENV_SOILNH4 |
| n_estimators | 60 |
| n_splits | 3 |
| test_size | 0.25 |
| run_univariate | True |
| run_permutation | True |
| run_shap | True |
| permutation_repeats | 10 |
| fdr_threshold | 0.05 |
Command history
# python from sklearn.ensemble import RandomForestRegressor # python model = RandomForestRegressor(n_estimators=60, random_state=42) # python scores = sklearn.model_selection.cross_val_score(model, X, y, cv=3)