10-fold on homework2

5 views
Skip to first unread message

Jonathan Daugaard

unread,
Feb 5, 2012, 11:08:04 AM2/5/12
to ml_chalmers
Hi

Are we supposed to do 10-fold cross validation when we report
classification erros on problem 1 and 2? If not which data should be
used for testing and which for training?

regards
Jonathan

Vinay Jethava

unread,
Feb 6, 2012, 6:01:52 AM2/6/12
to ml_ch...@googlegroups.com
No,

You use all the data for training and learn a classifier. Then, test this classifier on the same data that was used to train. This is similar to the setup for perceptron.

Vinay

Reply all
Reply to author
Forward
This conversation is locked
You cannot reply and perform actions on locked conversations.
0 new messages