ctr 平滑

spark/featureEng.py

@@ -229,13 +229,13 @@ def wilson_ctr(num_pv, num_click):
     num_pv = float(num_pv)
     num_click = float(num_click)
     if num_pv * num_click == 0 or num_pv < num_click:
-        return 0
+        return 0.0
 
     z = 1.96;
     n = num_pv;
     p = num_click / num_pv;
     score = (p + z*z/(2*n) - z*math.sqrt((p*(1.0 - p) + z*z /(4.0*n))/n)) / (1.0 + z*z/n);
-    return score;
+    return float(score);
 
 def addUserFeatures(samples,dataVocab,multiVocab):
     dataVocab["userid"] = collectColumnToVocab(samples, "userid")