From 4d2d6c91fcd1eb8214ac3c9e5ec8a2c07d0d0374 Mon Sep 17 00:00:00 2001
From: Kevin Crosby <Kevin@resolvity.com>
Date: Tue, 31 Jan 2017 12:05:20 -0600
Subject: [PATCH] Added log prior calculations in CachingNaiveBayesClassifier.

---
 .../classification/CachingNaiveBayesClassifier.java       | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/lucene/classification/src/java/org/apache/lucene/classification/CachingNaiveBayesClassifier.java b/lucene/classification/src/java/org/apache/lucene/classification/CachingNaiveBayesClassifier.java
index 6fe683546d06..30a3a358358e 100644
--- a/lucene/classification/src/java/org/apache/lucene/classification/CachingNaiveBayesClassifier.java
+++ b/lucene/classification/src/java/org/apache/lucene/classification/CachingNaiveBayesClassifier.java
@@ -88,11 +88,17 @@ protected List<ClassificationResult<BytesRef>> assignClassNormalizedList(String
     return asignedClassesNorm;
   }
 
+  private double calculateLogPrior(BytesRef cclass) throws IOException {
+    Term term = new Term(this.classFieldName, cclass);
+    int docsWithC = indexReader.docFreq(term);
+    return Math.log((double) docsWithC) - Math.log(docsWithClassSize);
+  }
+
   private List<ClassificationResult<BytesRef>> calculateLogLikelihood(String[] tokenizedText) throws IOException {
     // initialize the return List
     ArrayList<ClassificationResult<BytesRef>> ret = new ArrayList<>();
     for (BytesRef cclass : cclasses) {
-      ClassificationResult<BytesRef> cr = new ClassificationResult<>(cclass, 0d);
+      ClassificationResult<BytesRef> cr = new ClassificationResult<>(cclass, calculateLogPrior(cclass));
       ret.add(cr);
     }
     // for each word