update docstring

Author: Matt Sokoloff

labelbox/data/annotation_types/metrics/confusion_matrix.py

@@ -23,7 +23,7 @@ class ConfusionMatrixMetric(BaseMetric):
     In the editor, this provides precision, recall, and f-scores.
     This should be used over multiple scalar metrics so that aggregations are accurate.
 
-    value should be a tuple representing:
+    Value should be a tuple representing:
       [True Positive Count, False Positive Count, True Negative Count, False Negative Count]
 
     aggregation cannot be adjusted for confusion matrix metrics.

labelbox/data/metrics/confusion_matrix/calculation.py

@@ -81,6 +81,8 @@ def classification_confusion_matrix(
 ) -> ConfusionMatrixMetricValue:
     """
     Computes iou score for all features with the same feature schema id.
+    Because these predictions and ground truths are already sorted by schema id,
+    there can only be one of each (or zero if the classification was not predicted or labeled).
 
     Args:
         ground_truths: List of ground truth classification annotations

labelbox/data/metrics/confusion_matrix/confusion_matrix.py

@@ -20,7 +20,7 @@ def confusion_matrix_metric(ground_truths: List[Union[
                             iou=0.5) -> List[ConfusionMatrixMetric]:
     """
     Computes confusion matrix metrics between two sets of annotations.
-    This will most commonly be used for data row level metrics.
+    These annotations should relate to the same data (image/video).
     On the front end these will be displayed as precision, recall, and f1 scores.
 
     Args:
@@ -53,6 +53,7 @@ def feature_confusion_matrix_metric(
 ) -> List[ConfusionMatrixMetric]:
     """
     Computes the confusion matrix metrics for each type of class in the list of annotations.
+    These annotations should relate to the same data (image/video).
     On the front end these will be displayed as precision, recall, and f1 scores.
 
     Args: