Generate a weighted_num_examples stat if a weight feature is specified.

caveness · tf-data-validation-team · commit 96dfcb1c02e7 · 2019-09-27T11:51:12.000-07:00
PiperOrigin-RevId: 271616964
diff --git a/RELEASE.md b/RELEASE.md
@@ -12,6 +12,8 @@
   the default slice (validation/inference will be done using the default slice
   statistics).
 * Avoid flattening null arrays.
+* Set `weighted_num_examples` field in the statistics proto if a weight feature
+  is specified.
 
 
 ## Breaking Changes
diff --git a/tensorflow_data_validation/statistics/stats_impl.py b/tensorflow_data_validation/statistics/stats_impl.py
@@ -412,6 +412,7 @@ def _update_example_and_missing_count(
           weighted_num_examples -
           common_stats.weighted_common_stats.num_non_missing)
   stats.num_examples = int(num_examples)
+  stats.weighted_num_examples = weighted_num_examples
 
 
 def _make_dataset_feature_statistics_list_proto(
diff --git a/tensorflow_data_validation/statistics/stats_impl_test.py b/tensorflow_data_validation/statistics/stats_impl_test.py
@@ -317,6 +317,7 @@ def add_input(self, accumulator, feature_path, feature_array):
             """
             datasets {
               num_examples: 2
+              weighted_num_examples: 3
               features {
                 path {
                   step: "a"
