numerically more stable variance; but I think this could still be muc…

…h better
kgyrtkirk · Aug 18, 2023 · 48019e6 · 48019e6
1 parent db916dc
commit 48019e6
Showing 1 changed file with 4 additions and 16 deletions.
diff --git a/...rc/main/java/org/apache/druid/query/aggregation/variance/VarianceAggregatorCollector.java b/...rc/main/java/org/apache/druid/query/aggregation/variance/VarianceAggregatorCollector.java
@@ -127,35 +127,23 @@ public VarianceAggregatorCollector(long count, double sum, double nvariance)
 
   public VarianceAggregatorCollector add(float v)
   {
-    count++;
-    sum += v;
-    if (count > 1) {
-      double t = count * v - sum;
-      nvariance += (t * t) / ((double) count * (count - 1));
-    }
-    return this;
+    return add((double)v);
   }
 
   public VarianceAggregatorCollector add(double v)
   {
     count++;
     sum += v;
     if (count > 1) {
-      double t = count * v - sum;
-      nvariance += (t * t) / ((double) count * (count - 1));
+      double t = v - sum / count;
+      nvariance += (t * t) * count / (count - 1);
     }
     return this;
   }
 
   public VarianceAggregatorCollector add(long v)
   {
-    count++;
-    sum += v;
-    if (count > 1) {
-      double t = count * v - sum;
-      nvariance += (t * t) / ((double) count * (count - 1));
-    }
-    return this;
+    return add((double)v);
   }
 
   @Nullable