geocif 0.2.29__tar.gz → 0.2.31__tar.gz

{geocif-0.2.29/geocif.egg-info → geocif-0.2.31}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geocif
-Version: 0.2.29
+Version: 0.2.31
 Summary: Models to visualize and forecast crop conditions and yields
 Home-page: https://ritviksahajpal.github.io/yield_forecasting/
 Author: Ritvik Sahajpal

{geocif-0.2.29 → geocif-0.2.31}/geocif/geocif.py

@@ -294,7 +294,7 @@ class Geocif:
                         X_train,
                         self.y_train,
                         cat_features=self.cat_features,
-                        verbose=True,
+                        verbose=False,
                     )
                 elif self.model_name in ["ngboost", "oblique", "tabpfn"]:
                     X_train = X_train.drop(

{geocif-0.2.29 → geocif-0.2.31}/geocif/ml/feature_selection.py

@@ -7,6 +7,7 @@ from collections import Counter
 from pathlib import Path
 import matplotlib.pyplot as plt
 import pandas as pd
+import seaborn as sns
 
 
 def are_all_features_non_eo(features):
@@ -79,6 +80,8 @@ def select_features(
     # --- multi-method ensemble -------------------------------
     if method == "multi":
         counter = Counter()
+        selections = {}
+
         models = ["BorutaPy", "mrmr"]
         # run three selectors and count feature picks
         for sub_m in models:
@@ -89,6 +92,7 @@ def select_features(
                 threshold_nan=threshold_nan,
                 threshold_unique=threshold_unique
             )
+            selections[sub_m] = set(feats)
             counter.update(feats)
 
         # union of all features
@@ -109,6 +113,21 @@ def select_features(
         fig.savefig(dir_output / f"feature_selection_frequency_{region}.png", dpi=300)
         plt.close(fig)
 
+        # build DataFrame: rows=features, cols=methods, 1 if selected
+        sel_df = pd.DataFrame(
+            {m: [1 if f in selections[m] else 0 for f in combined] for m in models},
+            index=combined
+        )
+
+        fig2, ax2 = plt.subplots(figsize=(10, max(6, len(combined) * 0.3)))
+        sns.heatmap(sel_df, annot=True, cbar=False, linewidths=0.5, ax=ax2)
+        ax2.set_xlabel("Method")
+        ax2.set_ylabel("Feature")
+        ax2.set_title("Which features each method selected")
+        plt.tight_layout()
+        fig2.savefig(dir_output / f"feature_selection/feature_selection_methods_{region}.png", dpi=300)
+        plt.close(fig2)
+
         return None, X_out, combined
 
     # define forest for methods that need it

{geocif-0.2.29 → geocif-0.2.31/geocif.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: geocif
-Version: 0.2.29
+Version: 0.2.31
 Summary: Models to visualize and forecast crop conditions and yields
 Home-page: https://ritviksahajpal.github.io/yield_forecasting/
 Author: Ritvik Sahajpal

{geocif-0.2.29 → geocif-0.2.31}/setup.py

@@ -50,6 +50,6 @@ setup(
     test_suite="tests",
     tests_require=test_requirements,
     url="https://ritviksahajpal.github.io/yield_forecasting/",
-    version="0.2.29",
+    version="0.2.31",
     zip_safe=False,
 )