[ https://issues.apache.org/jira/browse/MAHOUT-997?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13242903#comment-13242903 ]
Lance Norskog commented on MAHOUT-997: -------------------------------------- This is a general problem, not a splitData problem. Suggest you prestage your Mahout input files from your real files with script that ignores the first line. {{sed 2,\$p}} will do the trick. Most input data requires some kind of cleanup. > Make splitData smart enough to not consider a CSV header to be part of the > data > ------------------------------------------------------------------------------- > > Key: MAHOUT-997 > URL: https://issues.apache.org/jira/browse/MAHOUT-997 > Project: Mahout > Issue Type: Improvement > Components: Integration > Affects Versions: 0.6 > Environment: OS X > Reporter: Andrew Harbick > Priority: Minor > Fix For: 0.6 > > > If you do something like: > MAHOUT_LOCAL=1 $MAHOUT_HOME/bin/mahout splitDataset --input all.csv --output > split --trainingPercentage 0.9 --probePercentage 0.1 > The header row from your CSV will end up with 90% chance in your training > data and 10% chance in your evaluation data. To use a tool like > trainlogistic or runlogistic the header file is needed in both. > Perhaps add an argument to splitData to duplicate the header line? -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira