Change Summary
Last iteration: 2022-09-08 - Landsat LiDAR 1.1.0: More data = more better
Change Summary
- This version uses PCA to reduce the dimensionality of the data
- We used 99% variance to select PCs
- Document includes comparisons against Landsat-FIA test plots
LiDAR-AGB Test Pixel Results
| RF (ranger) | GBM (LightGBM) | Ensemble (model weighted) | Ensemble (RMSE weighted) | |
|---|---|---|---|---|
| %RMSE | 31.064 | 28.419 | 28.005 | 28.395 |
| RMSE | 43.744 | 40.020 | 39.436 | 39.986 |
| MAE | 34.744 | 30.994 | 30.356 | 31.217 |
| MBE | -0.444 | -0.050 | -0.344 | -0.237 |
| R2 | 0.724 | 0.757 | 0.764 | 0.763 |
Landsat-FIA Test Plot Results
| RF (ranger) | GBM (LightGBM) | Ensemble (model weighted) | Ensemble (RMSE weighted) | |
|---|---|---|---|---|
| %RMSE | 43.059 | 42.096 | 42.164 | 41.909 |
| RMSE | 59.152 | 57.830 | 57.923 | 57.572 |
| MAE | 46.503 | 44.709 | 44.711 | 44.942 |
| MBE | 5.999 | 2.430 | 1.858 | 4.122 |
| R2 | 0.255 | 0.291 | 0.288 | 0.291 |
Landsat-FIA (With LiDAR-identified Zeroes) Test Plot Results
| RF (ranger) | GBM (LightGBM) | Ensemble (model weighted) | Ensemble (RMSE weighted) | |
|---|---|---|---|---|
| %RMSE | 46.209 | 44.423 | 44.499 | 44.497 |
| RMSE | 58.397 | 56.140 | 56.236 | 56.234 |
| MAE | 45.790 | 42.869 | 42.910 | 43.661 |
| MBE | 8.738 | 3.976 | 3.580 | 6.233 |
| R2 | 0.411 | 0.447 | 0.444 | 0.448 |
LiDAR-AGB Pixel Distribution
summary(bind_rows(training, testing)$agb_mgha) Min. 1st Qu. Median Mean 3rd Qu. Max.
0.00721 70.68408 141.33738 140.87515 212.00599 279.53638 
Metadata
Ensembles
- RMSE-weighted model weights:
- Linear model weights:
Call:
lm(formula = agb_mgha ~ rf_pred * lgb_pred, data = k_fold_preds)
Residuals:
Min 1Q Median 3Q Max
-179.487 -24.874 0.308 25.569 165.644
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -2.040e+00 1.691e+00 -1.207 0.228
rf_pred 2.571e-01 2.301e-02 11.175 < 2e-16 ***
lgb_pred 6.532e-01 2.286e-02 28.573 < 2e-16 ***
rf_pred:lgb_pred 6.087e-04 9.522e-05 6.392 1.71e-10 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 39.67 on 9996 degrees of freedom
Multiple R-squared: 0.761, Adjusted R-squared: 0.7609
F-statistic: 1.061e+04 on 3 and 9996 DF, p-value: < 2.2e-16\(n\) and \(p\)
- observations
- 10000 training
- testing
- 111 predictors
- PC1, PC2, PC3, PC4, PC5, PC6, PC7, PC8, PC9, PC10, PC11, PC12, PC13, PC14, PC15, PC16, PC17, PC18, PC19, PC20, PC21, PC22, PC23, PC24, PC25, PC26, PC27, PC28, PC29, PC30, PC31, PC32, PC33, PC34, PC35, PC36, PC37, PC38, PC39, PC40, PC41, PC42, PC43, PC44, PC45, PC46, PC47, PC48, PC49, PC50, PC51, PC52, PC53, PC54, PC55, PC56, PC57, PC58, PC59, PC60, PC61, PC62, PC63, PC64, PC65, PC66, PC67, PC68, PC69, PC70, PC71, PC72, PC73, PC74, PC75, PC76, PC77, PC78, PC79, PC80, PC81, PC82, PC83, PC84, PC85, PC86, PC87, PC88, PC89, PC90, PC91, PC92, PC93, PC94, PC95, PC96, PC97, PC98, PC99, PC100, PC101, PC102, PC103, PC104, PC105, PC106, PC107, PC108, PC109, PC110, PC111
Component Models
- Tuning used 5-fold CV
- Final hyperparameters:
Random forest:
$num.trees
[1] 750
$mtry
[1] 48
$min.node.size
[1] 1
$sample.fraction
[1] 1
$replace
[1] FALSE
$formula
agb_mgha ~ .LGB:
$params
$params$num_leaves
[1] 16
$params$max_depth
[1] 4
$params$extra_trees
[1] TRUE
$params$min_data_in_leaf
[1] 10
$params$bagging_fraction
[1] 0.3
$params$bagging_freq
[1] 0
$params$feature_fraction
[1] 1
$params$min_data_in_bin
[1] 2
$params$lambda_l1
[1] 7
$params$lambda_l2
[1] 5
$params$learning_rate
[1] 0.1
$params$force_col_wise
[1] TRUE
$params$nrounds
[1] 1500