siluette

setup.py

@@ -29,7 +29,7 @@
 # Setup configuration
 setuptools.setup(
     name="simba-uw-tf-dev",
-    version="2.4.6",
+    version="2.4.8",
     author="Simon Nilsson, Jia Jie Choong, Sophia Hwang",
     author_email="[email protected]",
     description="Toolkit for computer classification and analysis of behaviors in experimental animals",

simba/SimBA.py

@@ -356,7 +356,7 @@ def activate(box, *args):
 
         label_behavior_frm = CreateLabelFrameWithIcon(parent=tab7, header="LABEL BEHAVIOR", icon_name=Keys.DOCUMENTATION.value, icon_link=Links.LABEL_BEHAVIOR.value)
         select_video_btn_new = SimbaButton(parent=label_behavior_frm, txt="Select video (create new video annotation)", img='label_blue', txt_clr='navy', cmd=select_labelling_video, cmd_kwargs={'config_path': lambda :self.config_path, 'threshold_dict': lambda: None, 'setting': lambda: "from_scratch", 'continuing': lambda: False}, thread=False)
-        select_video_btn_continue = SimbaButton(parent=label_behavior_frm, txt="Select video (continue existing video annotation)", img='label_yellow', txt_clr='darkgoldenrod', cmd=select_labelling_video, cmd_kwargs={'config_path': lambda: self.config_path, 'threshold_dict': lambda:None, 'setting': lambda:None, 'continuing': lambda:True}, thread=False)
+        select_video_btn_continue = SimbaButton(parent=label_behavior_frm, txt="Select video (continue existing video annotation)", img='label_yellow', txt_clr='darkgoldenrod', cmd=select_labelling_video, cmd_kwargs={'config_path': lambda: self.config_path, 'threshold_dict': lambda:None, 'setting': lambda: None, 'continuing': lambda:True}, thread=False)
 
         label_thirdpartyann = CreateLabelFrameWithIcon(parent=tab7, header="IMPORT THIRD-PARTY BEHAVIOR ANNOTATIONS", icon_name=Keys.DOCUMENTATION.value, icon_link=Links.THIRD_PARTY_ANNOTATION.value)
         button_importmars = SimbaButton(parent=label_thirdpartyann, txt="Import MARS Annotation (select folder with .annot files)", txt_clr="blue", cmd=self.importMARS, thread=False)

simba/labelling/labelling_interface.py

@@ -63,7 +63,7 @@ def __init__(self,
                  config_path: Union[str, os.PathLike],
                  file_path: Union[str, os.PathLike],
                  threshold_dict: Optional[Dict[str, float]] = None,
-                 setting: Literal["from_scratch", "pseudo"] = "pseudo",
+                 setting: Optional[Literal["from_scratch", "pseudo"]] = "from_scratch",
                  continuing: Optional[bool] = False):
 
         ConfigReader.__init__(self, config_path=config_path)
@@ -360,7 +360,8 @@ def select_labelling_video(config_path: Union[str, os.PathLike],
     check_file_exist_and_readable(file_path=config_path)
     if threshold_dict is not None:
         check_valid_dict(x=threshold_dict, valid_key_dtypes=(str,), valid_values_dtypes=(float,))
-    check_str(name='setting', value=setting, options=('pseudo', "from_scratch"))
+    if setting is not None:
+        check_str(name='setting', value=setting, options=('pseudo', "from_scratch",))
     check_valid_boolean(value=[continuing], source=select_labelling_video.__name__)
 
 

simba/mixins/statistics_mixin.py

@@ -3931,6 +3931,45 @@ def calinski_harabasz(x: np.ndarray, y: np.ndarray) -> float:
         else:
             return extra_dispersion * (x.shape[0] - n_labels) / denominator
 
+    def silhouette_score(self, x: np.ndarray, y: np.ndarray) -> float:
+        """
+        Compute the silhouette score for the given dataset and labels.
+
+        :param np.ndarray x: The dataset as a 2D NumPy array of shape (n_samples, n_features).
+        :param np.ndarray y: Cluster labels for each data point as a 1D NumPy array of shape (n_samples,).
+        :returns: The average silhouette score for the dataset.
+        :rtype: float
+
+        :example:
+        >>> x, y = make_blobs(n_samples=10000, n_features=400, centers=5, cluster_std=10, center_box=(-1, 1))
+        >>> score = silhouette_score(x=x, y=y)
+
+        >>> from sklearn.metrics import silhouette_score as sklearn_silhouette # SKLEARN ALTERNATIVE
+        >>> score_sklearn = sklearn_silhouette(x, y)
+
+        """
+        dists = cdist(x, x)
+        results = np.full(x.shape[0], fill_value=-1.0, dtype=np.float32)
+        cluster_ids = np.unique(y)
+        cluster_indices = {cluster_id: np.argwhere(y == cluster_id).flatten() for cluster_id in cluster_ids}
+
+        for i in range(x.shape[0]):
+            intra_idx = cluster_indices[y[i]]
+            if len(intra_idx) <= 1:
+                a_i = 0.0
+            else:
+                intra_distances = dists[i, intra_idx]
+                a_i = np.sum(intra_distances) / (intra_distances.shape[0] - 1)
+            b_i = np.inf
+            for cluster_id in cluster_ids:
+                if cluster_id != y[i]:
+                    inter_idx = cluster_indices[cluster_id]
+                    inter_distances = dists[i, inter_idx]
+                    b_i = min(b_i, np.mean(inter_distances))
+            results[i] = (b_i - a_i) / max(a_i, b_i)
+
+        return np.mean(results)
+
     @staticmethod
     def adjusted_rand(x: np.ndarray, y: np.ndarray) -> float:
         """