nkululeko 0.95.5__tar.gz → 0.95.7__tar.gz

{nkululeko-0.95.5 → nkululeko-0.95.7}/CHANGELOG.md

@@ -1,6 +1,15 @@
 Changelog
 =========
 
+Version 0.95.7 (25-07-11)
+--------------------------
+* infer gender and age from audb misc table
+* fix result plots in run folder
+
+Version 0.95.6 (25-07-10)
+--------------------------
+* added uncertainty_threshold plot
+
 Version 0.95.5 (25-07-08)
 --------------------------
 * fixed bug that categorical targets couldn't be super mapped

{nkululeko-0.95.5/nkululeko.egg-info → nkululeko-0.95.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nkululeko
-Version: 0.95.5
+Version: 0.95.7
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt

{nkululeko-0.95.5 → nkululeko-0.95.7}/nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.95.5"
+VERSION="0.95.7"
 SAMPLING_RATE = 16000

{nkululeko-0.95.5 → nkululeko-0.95.7}/nkululeko/data/dataset.py

@@ -294,12 +294,21 @@ class Dataset:
                 # try to get the gender values
                 if "gender" in source_df:
                     df_local["gender"] = source_df["gender"]
-                    got_gender = True
+                else:
+                    # try to get the gender via the speaker description
+                    gender_map = db['speaker'].get().to_dict()['gender']
+                    df_local['gender'] = df_local['speaker'].map(gender_map).astype(str)
+                got_gender = True
             except (KeyError, ValueError, audformat.errors.BadKeyError):
                 pass
             try:
                 # try to get the age values
-                df_local["age"] = source_df["age"].astype(int)
+                if "age" in source_df:
+                    df_local["age"] = source_df["age"].astype(int)
+                else:
+                    # try to get the age via the speaker description
+                    age_map = db['speaker'].get().to_dict()['age']
+                    df_local['age'] = df_local['speaker'].map(age_map).astype(int)
                 got_age = True
             except (KeyError, ValueError, audformat.errors.BadKeyError):
                 pass

{nkululeko-0.95.5 → nkululeko-0.95.7}/nkululeko/explore.py

@@ -65,6 +65,7 @@ def main():
     try:
         # load the experiment
         expr.load(f"{util.get_save_name()}")
+        expr.util.set_config(config)
         needs_feats = True
         experiment_loaded = True
     except FileNotFoundError:

{nkululeko-0.95.5 → nkululeko-0.95.7}/nkululeko/plots.py

@@ -9,6 +9,7 @@ from scipy import stats
 import seaborn as sns
 from sklearn.manifold import TSNE
 
+import audeer
 from audmetric import concordance_cc as ccc
 
 import nkululeko.glob_conf as glob_conf
@@ -218,7 +219,7 @@ class Plots:
 
     def save_plot(self, ax, caption, header, filename, type_s):
         # one up because of the runs
-        fig_dir = os.path.dirname(self.util.get_path("fig_dir"))
+        fig_dir = audeer.path(self.util.get_path("fig_dir"), "..")
         fig_plots = ax.figure
         # avoid warning
         # plt.tight_layout()

{nkululeko-0.95.5 → nkululeko-0.95.7}/nkululeko/reporting/reporter.py

@@ -12,13 +12,9 @@ from scipy.special import softmax
 from scipy.stats import entropy
 from scipy.stats import pearsonr
 from sklearn.metrics import ConfusionMatrixDisplay
-from sklearn.metrics import RocCurveDisplay
-from sklearn.metrics import auc
 from sklearn.metrics import classification_report
 from sklearn.metrics import confusion_matrix
 from sklearn.metrics import r2_score
-from sklearn.metrics import roc_auc_score
-from sklearn.metrics import roc_curve
 
 # from torch import is_tensor
 from audmetric import accuracy
@@ -186,6 +182,7 @@ class Reporter:
                     if not file_name.endswith(".csv"):
                         file_name = file_name + ".csv"
             self.probas = probas
+            self.plot_proba_conf()
             probas.to_csv(file_name)
             self.util.debug(f"Saved probabilities to {file_name}")
             plots = Plots()
@@ -196,10 +193,27 @@ class Reporter:
                 ax,
                 caption,
                 "Uncertainty",
-                "uncertainty_samples",
+                "uncertainty",
                 "samples",
             )
 
+    def plot_proba_conf(self):
+        uncertainty_threshold = self.util.config_val("PLOT", "uncertainty_threshold", False)
+        if uncertainty_threshold:
+            uncertainty_threshold = float(uncertainty_threshold)
+            old_size = self.probas.shape[0]
+            df = self.probas[self.probas["uncertainty"] < uncertainty_threshold]
+            new_size = df.shape[0]
+            difference = old_size - new_size
+            self.util.debug(
+                f"Filtered probabilities: {old_size} -> {new_size} ({difference}) samples with uncertainty < {uncertainty_threshold}"
+            )
+            truths = df["truth"].values
+            preds = df["predicted"].values
+            self._plot_confmat(truths, preds, f"uncertainty_less_than_{uncertainty_threshold}_cnf",
+                            epoch=None, test_result=None)
+
+
     def set_id(self, run, epoch):
         """Make the report identifiable with run and epoch index."""
         self.run = run

{nkululeko-0.95.5 → nkululeko-0.95.7/nkululeko.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nkululeko
-Version: 0.95.5
+Version: 0.95.7
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt

{nkululeko-0.95.5 → nkululeko-0.95.7}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "nkululeko"
-version = "0.95.5"
+version = "0.95.7"
 description = "Machine learning audio prediction experiments based on templates"
 authors = [
     {name = "Felix Burkhardt", email = "fxburk@gmail.com"},