kz930 commented on PR #7668:
URL: https://github.com/apache/texera/pull/7668#issuecomment-5392205564

   Tested with a real workflow. Two CSV File Scans feed the training and 
testing ports of Logistic Regression, with `label` as target and Count 
Vectorizer on. The data has its spam signal split across two text columns: for 
half the rows the giveaway word is in `title`, for the other half in `body`, 
and the two titles that repeat appear once as spam and once as ham so `title` 
alone carries no signal for them.
   
   Naming only `title` gives Overall Accuracy 0.75, with spam recall 0.5, 
missing exactly the test row whose signal is in `body`. Naming `title` and 
`body` gives 1.0 on all four test rows. Same run also covers the single-column 
case, which is the path a pre-existing workflow takes.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to