David Engel created MAHOUT-1062:
-----------------------------------
Summary: alphaI is not correctly saved in NaiveBayesModel
Key: MAHOUT-1062
URL: https://issues.apache.org/jira/browse/MAHOUT-1062
Project: Mahout
Issue Type: Bug
Components: Classification
Affects Versions: 0.7
Reporter: David Engel
alphaI is passed in for training via the thetaSummer job configuration. When
the model is saved, however, the wrong configuration is used causing the saved
alphaI to always be 1.0. The following patch fixes the problem for me.
diff -ur
mahout-distribution-0.7.orig/core/src/main/java/org/apache/mahout/classifier/naivebayes/training/TrainNaiveBayesJob.java
mahout-distribution-0.7/core/src/main/java/org/apache/mahout/classifier/naivebayes/training/TrainNaiveBayesJob.java
---
mahout-distribution-0.7.orig/core/src/main/java/org/apache/mahout/classifier/naivebayes/training/TrainNaiveBayesJob.java
2012-06-12 03:32:17.000000000 -0500
+++
mahout-distribution-0.7/core/src/main/java/org/apache/mahout/classifier/naivebayes/training/TrainNaiveBayesJob.java
2012-08-28 13:06:11.000000000 -0500
@@ -134,7 +134,7 @@
}*/
//validate our model and then write it out to the official output
- NaiveBayesModel naiveBayesModel =
BayesUtils.readModelFromDir(getTempPath(), getConf());
+ NaiveBayesModel naiveBayesModel =
BayesUtils.readModelFromDir(getTempPath(), thetaSummer.getConfiguration());
naiveBayesModel.validate();
naiveBayesModel.serialize(getOutputPath(), getConf());
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
