Author: srowen
Date: Sat Jan 10 09:09:52 2009
New Revision: 733301
URL: http://svn.apache.org/viewvc?rev=733301&view=rev
Log:
SequenceFileModelReader can/should be a utility class
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/Classify.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/BayesThetaNormalizerDriver.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesTfIdfDriver.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/io/SequenceFileModelReader.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesNormalizedWeightDriver.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaDriver.java
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaNormalizerDriver.java
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/Classify.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/Classify.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/Classify.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/Classify.java
Sat Jan 10 09:09:52 2009
@@ -96,7 +96,6 @@
parser.setGroup(options);
CommandLine cmdLine = parser.parse(args);
- SequenceFileModelReader reader = new SequenceFileModelReader();
JobConf conf = new JobConf(Classify.class);
Map<String, Path> modelPaths = new HashMap<String, Path>();
@@ -126,7 +125,7 @@
}
Classifier classifier = new BayesClassifier();
- reader.loadModel(model, fs, modelPaths, conf);
+ SequenceFileModelReader.loadModel(model, fs, modelPaths, conf);
log.info("Done loading model: # labels: {}", model.getLabels().size());
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/BayesThetaNormalizerDriver.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/BayesThetaNormalizerDriver.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/BayesThetaNormalizerDriver.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/BayesThetaNormalizerDriver.java
Sat Jan 10 09:09:52 2009
@@ -93,8 +93,6 @@
if (dfs.exists(outPath))
dfs.delete(outPath, true);
- SequenceFileModelReader reader = new SequenceFileModelReader();
-
Path Sigma_kFiles = new Path(output+"/trainer-weights/Sigma_k/*");
Map<String,Double> labelWeightSum =
SequenceFileModelReader.readLabelSums(dfs, Sigma_kFiles, conf);
DefaultStringifier<Map<String,Double>> mapStringifier =
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesTfIdfDriver.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesTfIdfDriver.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesTfIdfDriver.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/common/BayesTfIdfDriver.java
Sat Jan 10 09:09:52 2009
@@ -95,8 +95,6 @@
if (dfs.exists(outPath))
dfs.delete(outPath, true);
- SequenceFileModelReader reader = new SequenceFileModelReader();
-
Path interimFile = new Path(output+"/trainer-docCount/part-*");
Map<String,Double> labelDocumentCounts =
SequenceFileModelReader.readLabelDocumentCounts(dfs, interimFile, conf);
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/io/SequenceFileModelReader.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/io/SequenceFileModelReader.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/io/SequenceFileModelReader.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/bayes/io/SequenceFileModelReader.java
Sat Jan 10 09:09:52 2009
@@ -40,7 +40,9 @@
*/
public class SequenceFileModelReader {
- private static final Logger log =
LoggerFactory.getLogger(SequenceFileModelReader.class);
+ private static final Logger log =
LoggerFactory.getLogger(SequenceFileModelReader.class);
+
+ private SequenceFileModelReader() {}
public static void loadModel(Model model, FileSystem fs, Map<String, Path>
pathPatterns,
Configuration conf) throws IOException {
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesNormalizedWeightDriver.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesNormalizedWeightDriver.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesNormalizedWeightDriver.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesNormalizedWeightDriver.java
Sat Jan 10 09:09:52 2009
@@ -93,8 +93,6 @@
if (dfs.exists(outPath))
dfs.delete(outPath, true);
- SequenceFileModelReader reader = new SequenceFileModelReader();
-
Path thetaNormalizationsFiles = new
Path(output+"/trainer-thetaNormalizer/part*");
Map<String,Double> thetaNormalizer=
SequenceFileModelReader.readLabelSums(dfs, thetaNormalizationsFiles, conf);
double perLabelWeightSumNormalisationFactor = Double.MAX_VALUE;
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaDriver.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaDriver.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaDriver.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaDriver.java
Sat Jan 10 09:09:52 2009
@@ -95,8 +95,6 @@
if (dfs.exists(outPath))
dfs.delete(outPath, true);
- SequenceFileModelReader reader = new SequenceFileModelReader();
-
Path Sigma_kFiles = new Path(output+"/trainer-weights/Sigma_k/*");
Map<String,Double> labelWeightSum=
SequenceFileModelReader.readLabelSums(dfs, Sigma_kFiles, conf);
DefaultStringifier<Map<String,Double>> mapStringifier =
Modified:
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaNormalizerDriver.java
URL:
http://svn.apache.org/viewvc/lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaNormalizerDriver.java?rev=733301&r1=733300&r2=733301&view=diff
==============================================================================
---
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaNormalizerDriver.java
(original)
+++
lucene/mahout/trunk/core/src/main/java/org/apache/mahout/classifier/cbayes/CBayesThetaNormalizerDriver.java
Sat Jan 10 09:09:52 2009
@@ -94,8 +94,6 @@
if (dfs.exists(outPath))
dfs.delete(outPath, true);
- SequenceFileModelReader reader = new SequenceFileModelReader();
-
Path Sigma_kFiles = new Path(output+"/trainer-weights/Sigma_k/*");
Map<String,Double> labelWeightSum=
SequenceFileModelReader.readLabelSums(dfs, Sigma_kFiles, conf);
DefaultStringifier<Map<String,Double>> mapStringifier =