Author: srowen
Date: Sun Nov 9 09:11:00 2008
New Revision: 712520
URL: http://svn.apache.org/viewvc?rev=712520&view=rev
Log:
Tiny performance improvements / streamlining
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/cf/taste/hadoop/SlopeOnePrefsToDiffsReducer.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/BayesFileFormatter.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesFeatureMapper.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/common/Model.java
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/cf/taste/hadoop/SlopeOnePrefsToDiffsReducer.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/cf/taste/hadoop/SlopeOnePrefsToDiffsReducer.java?rev=712520&r1=712519&r2=712520&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/cf/taste/hadoop/SlopeOnePrefsToDiffsReducer.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/cf/taste/hadoop/SlopeOnePrefsToDiffsReducer.java
Sun Nov 9 09:11:00 2008
@@ -27,7 +27,6 @@
import java.io.IOException;
import java.util.ArrayList;
import java.util.Collections;
-import java.util.Comparator;
import java.util.Iterator;
import java.util.List;
@@ -43,7 +42,7 @@
while (values.hasNext()) {
prefs.add(new ItemPrefWritable(values.next()));
}
- Collections.sort(prefs, (Comparator<? super ItemPrefWritable>)
ByItemIDComparator.getInstance());
+ Collections.sort(prefs, ByItemIDComparator.getInstance());
int size = prefs.size();
for (int i = 0; i < size; i++) {
ItemPrefWritable first = prefs.get(i);
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/BayesFileFormatter.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/BayesFileFormatter.java?rev=712520&r1=712519&r2=712520&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/BayesFileFormatter.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/BayesFileFormatter.java
Sun Nov 9 09:11:00 2008
@@ -204,8 +204,8 @@
writer.write(termBuffer, 0, termLen);
writer.write(' ');
- char[] tmp = new char[termLen];
- System.arraycopy(termBuffer, 0, tmp, 0, termLen);
+ //char[] tmp = new char[termLen];
+ //System.arraycopy(termBuffer, 0, tmp, 0, termLen);
//seen.add(tmp);// do this b/c CharArraySet doesn't allow offsets
}
///numTokens++;
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesFeatureMapper.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesFeatureMapper.java?rev=712520&r1=712519&r2=712520&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesFeatureMapper.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesFeatureMapper.java
Sun Nov 9 09:11:00 2008
@@ -70,31 +70,32 @@
String label = key.toString();
int keyLen = label.length();
- Map<String, Integer[]> wordList = new HashMap<String, Integer[]>(1000);
+ Map<String, int[]> wordList = new HashMap<String, int[]>(1000);
StringBuilder builder = new StringBuilder(label);
builder.ensureCapacity(32);// make sure we have a reasonably size buffer to
// begin with
List<String> ngrams = Model.generateNGramsWithoutLabel(value.toString(),
gramSize);
for (String ngram : ngrams) {
- Integer[] count = wordList.get(ngram);
+ int[] count = wordList.get(ngram);
if (count == null) {
- count = new Integer[1];
+ count = new int[1];
count[0] = 0;
wordList.put(ngram, count);
}
count[0]++;
}
double lengthNormalisation = 0.0;
- for (Integer[] D_kj : wordList.values()) {
+ for (int[] D_kj : wordList.values()) {
// key is label,word
- lengthNormalisation += D_kj[0].doubleValue() * D_kj[0].doubleValue();
+ double dkjValue = (double) D_kj[0];
+ lengthNormalisation += dkjValue * dkjValue;
}
lengthNormalisation = Math.sqrt(lengthNormalisation);
// Output Length Normalized + TF Transformed Frequency per Word per Class
// Log(1 + D_ij)/SQRT( SIGMA(k, D_kj) )
- for (Map.Entry<String, Integer[]> entry : wordList.entrySet()) {
+ for (Map.Entry<String, int[]> entry : wordList.entrySet()) {
// key is label,word
String token = entry.getKey();
builder.append(',').append(token);
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/common/Model.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/common/Model.java?rev=712520&r1=712519&r2=712520&view=diff
==============================================================================
--- lucene/mahout/trunk/core/src/main/java/org/apache/mahout/common/Model.java
(original)
+++ lucene/mahout/trunk/core/src/main/java/org/apache/mahout/common/Model.java
Sun Nov 9 09:11:00 2008
@@ -83,15 +83,16 @@
if (!labelList.containsKey(label)) {
Integer labelId = labelList.size();
labelList.put(label, labelId);
+ return labelId;
}
return labelList.get(label);
}
protected Integer getFeature(String feature) {
if (!featureList.containsKey(feature)) {
-
Integer featureId = featureList.size();
featureList.put(feature, featureId);
+ return featureId;
}
return featureList.get(feature);
}
