From 282d2991877df6d117e0e6377ae2486f0b3d4ccb Mon Sep 17 00:00:00 2001
From: ManuelKurmann <147698793+ManuelKurmann@users.noreply.github.com>
Date: Mon, 27 Apr 2026 14:31:41 +0200
Subject: [PATCH 1/2] Feedback on testing the tutorial in evoland.qmd

---
 vignettes/evoland.qmd | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/vignettes/evoland.qmd b/vignettes/evoland.qmd
index ea6121e..df5447e 100644
--- a/vignettes/evoland.qmd
+++ b/vignettes/evoland.qmd
@@ -370,7 +370,7 @@ db$trans_preds_t <- db$get_pruned_trans_preds_t(
 
 Now we fit partial models using training/validation splits, allowing for a goodness-of-fit (gof) estimation.
 Here, we only fit a single partial model per transition, where we normally would create a list of candidates.
-We can then pick the models with the best goodness of fit to retrain full models on all of the available predictor data; the full models are then used during extrapolation.
+Next, fit_full_models() reads the partial models stored in db$trans_models_t, chooses the best model for each transition based on goodness of fit, and refits each chosen model on all available predictor data. Assigning the result back to db$trans_models_t stores these full models for the extrapolation step.
 
 ```{r}
 #| label: trans-models
@@ -419,6 +419,8 @@ With all components in place, we run the allocation step.
 If Dinamica is not installed, you'll get a warning that the anterior LULC map is returned as the posterior.
 For installation instructions, see the [Installing Dinamica EGO](install-dinamica.html) vignette.
 
+<!-- TODO: Check for a possible error here. In my run, db$alloc_dinamica(...) failed when writing the transition probability maps with: unable to find an inherited method for function 'rasterize' for signature '"data.table", "SpatRaster"'. db$predict_trans_pot() returned a trans_pot_t/data.table with id_trans, id_period_post, id_coord, and value, could the issue be in the table-to-raster conversion inside the allocation workflow? -->
+
 ```{r}
 #| label: allocation
 #| results: hold

From 2572858ff8a4b8cbdfe7ba56e51f4c7b8e77299f Mon Sep 17 00:00:00 2001
From: ManuelKurmann <147698793+ManuelKurmann@users.noreply.github.com>
Date: Mon, 27 Apr 2026 15:31:30 +0200
Subject: [PATCH 2/2] Small fixes

- Bump terra to >= 1.8-42 for data.table rasterization support
- Pass GOF selection args to alloc_dinamica() in test_integ_allocation.R.
- TODO removed
---
 DESCRIPTION                           | 2 +-
 inst/tinytest/test_integ_allocation.R | 2 ++
 vignettes/evoland.qmd                 | 2 --
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/DESCRIPTION b/DESCRIPTION
index 322f390..e62abf1 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -22,7 +22,7 @@ Imports:
     qs2,
     R6,
     Rcpp,
-    terra
+    terra (>=1.8-42)
 Suggests:
     base64enc,
     butcher,
diff --git a/inst/tinytest/test_integ_allocation.R b/inst/tinytest/test_integ_allocation.R
index 06884d7..60e9fc1 100644
--- a/inst/tinytest/test_integ_allocation.R
+++ b/inst/tinytest/test_integ_allocation.R
@@ -49,6 +49,8 @@ if (Sys.which("DinamicaConsole") == "") {
   expect_stdout(
     db$alloc_dinamica(
       id_periods = 4,
+      gof_criterion = "auc",
+      gof_maximize = TRUE,
       work_dir = file.path(db$path, "dinamica_test"),
       keep_intermediate = FALSE
     ),
diff --git a/vignettes/evoland.qmd b/vignettes/evoland.qmd
index df5447e..bd005e0 100644
--- a/vignettes/evoland.qmd
+++ b/vignettes/evoland.qmd
@@ -419,8 +419,6 @@ With all components in place, we run the allocation step.
 If Dinamica is not installed, you'll get a warning that the anterior LULC map is returned as the posterior.
 For installation instructions, see the [Installing Dinamica EGO](install-dinamica.html) vignette.
 
-<!-- TODO: Check for a possible error here. In my run, db$alloc_dinamica(...) failed when writing the transition probability maps with: unable to find an inherited method for function 'rasterize' for signature '"data.table", "SpatRaster"'. db$predict_trans_pot() returned a trans_pot_t/data.table with id_trans, id_period_post, id_coord, and value, could the issue be in the table-to-raster conversion inside the allocation workflow? -->
-
 ```{r}
 #| label: allocation
 #| results: hold