nkululeko 0.96.0__tar.gz → 0.96.2__tar.gz

{nkululeko-0.96.0 → nkululeko-0.96.2}/CHANGELOG.md

@@ -1,6 +1,15 @@
 Changelog
 =========
 
+Version 0.96.2 (25-07-16)
+--------------------------
+* bugfix: scatter plot for regression problems was not generated any more
+* add category counts to console output
+
+Version 0.96.1 (25-07-16)
+--------------------------
+* bugfix: wrong labels in confmatrix plots
+
 Version 0.96.0 (25-07-14)
 --------------------------
 * added Bert feature extractor

{nkululeko-0.96.0/nkululeko.egg-info → nkululeko-0.96.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nkululeko
-Version: 0.96.0
+Version: 0.96.2
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt

{nkululeko-0.96.0 → nkululeko-0.96.2}/nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.96.0"
+VERSION="0.96.2"
 SAMPLING_RATE = 16000

{nkululeko-0.96.0 → nkululeko-0.96.2}/nkululeko/experiment.py

@@ -300,43 +300,34 @@ class Experiment:
             else:
                 if not self.test_empty:
                     if self.df_test.is_labeled:
-                        test_cats = self.df_test[self.target].unique()
+                        # get printable string of categories and their counts
+                        test_cats = self.df_test[self.target].value_counts().to_string()
                     else:
                         # if there is no target, copy a dummy label
                         self.df_test = self._add_random_target(self.df_test).astype(
                             "str"
                         )
                 if not self.train_empty:
-                    train_cats = self.df_train[self.target].unique()
+                    train_cats = self.df_train[self.target].value_counts().to_string()
                 if self.split3 and not self.dev_empty:
-                    dev_cats = self.df_dev[self.target].unique()
+                    dev_cats = self.df_dev[self.target].value_counts().to_string()
             # encode the labels as numbers
             self.label_encoder = LabelEncoder()
             glob_conf.set_label_encoder(self.label_encoder)
             if not self.train_empty:
-                if isinstance(train_cats, np.ndarray):
-                    self.util.debug(f"Categories train (nd.array): {train_cats}")
-                else:
-                    self.util.debug(f"Categories train (list): {list(train_cats)}")
-
+                self.util.debug(f"Categories train: {train_cats}")
                 self.df_train[self.target] = self.label_encoder.fit_transform(
                     self.df_train[self.target]
                 )
             if not self.test_empty:
                 if self.df_test.is_labeled:
-                    if isinstance(test_cats, np.ndarray):
-                        self.util.debug(f"Categories test (nd.array): {test_cats}")
-                    else:
-                        self.util.debug(f"Categories test (list): {list(test_cats)}")
+                    self.util.debug(f"Categories test: {test_cats}")
                 if not self.train_empty:
                     self.df_test[self.target] = self.label_encoder.transform(
                         self.df_test[self.target]
                     )
             if self.split3 and not self.dev_empty:
-                if isinstance(dev_cats, np.ndarray):
-                    self.util.debug(f"Categories dev (nd.array): {dev_cats}")
-                else:
-                    self.util.debug(f"Categories dev (list): {list(dev_cats)}")
+                self.util.debug(f"Categories dev: {dev_cats}")
                 if not self.train_empty:
                     self.df_dev[self.target] = self.label_encoder.transform(
                         self.df_dev[self.target]

{nkululeko-0.96.0 → nkululeko-0.96.2}/nkululeko/reporting/reporter.py

@@ -198,7 +198,9 @@ class Reporter:
             )
 
     def plot_proba_conf(self):
-        uncertainty_threshold = self.util.config_val("PLOT", "uncertainty_threshold", False)
+        uncertainty_threshold = self.util.config_val(
+            "PLOT", "uncertainty_threshold", False
+        )
         if uncertainty_threshold:
             uncertainty_threshold = float(uncertainty_threshold)
             old_size = self.probas.shape[0]
@@ -210,9 +212,13 @@ class Reporter:
             )
             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)
-
+            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."""
@@ -235,9 +241,7 @@ class Reporter:
             epoch (int, optional): Number of epoch. Defaults to None.
         """
         if not self.util.exp_is_classification():
-            self._plot_scatter(
-                self.truths, self.preds, plot_name.replace("cnf", "scatter"), epoch
-            )
+            self._plot_scatter(self.truths, self.preds, f"{plot_name}_scatter", epoch)
             self.continuous_to_categorical()
         self._plot_confmat(self.truths, self.preds, plot_name, epoch)
 
@@ -434,7 +438,10 @@ class Reporter:
             self.util.debug(f"####->{file_name}<-####")
             file_name = f"{res_dir}{file_name}{self.filenameadd}.txt"
         if self.util.exp_is_classification():
-            labels = glob_conf.labels
+            if glob_conf.label_encoder is not None:
+                labels = glob_conf.label_encoder.classes_
+            else:
+                labels = glob_conf.labels
             try:
                 rpt = classification_report(
                     self.truths,
@@ -451,9 +458,7 @@ class Reporter:
                     target_names=s_labels,
                     digits=4,
                 )
-                self.util.debug(
-                    f"\n {class_report_str}"
-                )
+                self.util.debug(f"\n {class_report_str}")
             except ValueError as e:
                 self.util.debug(
                     "Reporter: caught a ValueError when trying to get"

{nkululeko-0.96.0 → nkululeko-0.96.2/nkululeko.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nkululeko
-Version: 0.96.0
+Version: 0.96.2
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt

{nkululeko-0.96.0 → nkululeko-0.96.2}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "nkululeko"
-version = "0.96.0"
+version = "0.96.2"
 description = "Machine learning audio prediction experiments based on templates"
 authors = [
     {name = "Felix Burkhardt", email = "fxburk@gmail.com"},