Merge pull request #1691 from microsoft/andreas/exp_var

Use numpy divide in explained variance
recommenders-team · Mar 31, 2022 · f52cef2 · f52cef2
2 parents 7721857 + 75485fe
commit f52cef2
Show file tree

Hide file tree

Showing 3 changed files with 5 additions and 3 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -1,6 +1,6 @@
 # What's New
 
-## Update March 31, 2022
+## Update April 1, 2022
 
 We have a new release [Recommenders 1.1.0](https://github.com/microsoft/recommenders/releases/tag/1.1.0)! 
 We have introduced the SASRec and SSEPT algorithms that are based on transformers. 

diff --git a/README.md b/README.md
@@ -2,7 +2,7 @@
 
 [![Documentation Status](https://readthedocs.org/projects/microsoft-recommenders/badge/?version=latest)](https://microsoft-recommenders.readthedocs.io/en/latest/?badge=latest)
 
-## What's New (March 31, 2022)
+## What's New (April 1, 2022)
 
 We have a new release [Recommenders 1.1.0](https://github.com/microsoft/recommenders/releases/tag/1.1.0)! 
 We have introduced the SASRec and SSEPT algorithms that are based on transformers. 

diff --git a/recommenders/evaluation/spark_evaluation.py b/recommenders/evaluation/spark_evaluation.py
@@ -1,6 +1,7 @@
 # Copyright (c) Microsoft Corporation. All rights reserved.
 # Licensed under the MIT License.
 
+import numpy as np
 try:
     from pyspark.mllib.evaluation import RegressionMetrics, RankingMetrics
     from pyspark.sql import Window, DataFrame
@@ -158,7 +159,8 @@ def exp_var(self):
             0
         ]
         var2 = self.y_pred_true.selectExpr("variance(label)").collect()[0][0]
-        return 1 - var1 / var2
+        # numpy divide is more tolerant to var2 being zero
+        return 1 - np.divide(var1, var2)
 
 
 class SparkRankingEvaluation: