homework 3.4.1 test/train error

[Jingkun Gao]

What's the difference between test error and train error?

I thought there is only one kind of error. For example, in N-fold cv, each cv will generate an error, which is calculated using wrong labelled test samples divide total test samples. We just sum up the errors in the loop and divided by N. What's the name of this error? 

Thanks,

Regards,

Jingkun

[Vagelis Papalexakis]

That's the test error.

For the train error you do the cross validation using the train data (i.e. the same data that you used to train the model). In other words, you train and test based on the same piece of data.

You usually want to see this because you want to see how much your model is overfitted with respect to the train data.

Hope that helps,

Vagelis

--
http://alex.smola.org/teaching/cmu2013-10-701 (course website)
http://www.youtube.com/playlist?list=PLZSO_6-bSqHQmMKwWVvYwKreGu4b4kMU9 (YouTube playlist)
---
You received this message because you are subscribed to the Google Groups "10-701 Spring 2013 CMU" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 10-701-spring-201...@googlegroups.com.
To post to this group, send email to 10-701-spri...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Jingkun Gao]

Got it.

That's really helpful. Thanks a lot.

Regards,

Jingkun

On Tuesday, February 5, 2013 9:34:20 PM UTC-5, Vagelis Papalexakis wrote:

That's the test error.

For the train error you do the cross validation using the train data (i.e. the same data that you used to train the model). In other words, you train and test based on the same piece of data.

You usually want to see this because you want to see how much your model is overfitted with respect to the train data.

Hope that helps,

Vagelis

On Feb 5, 2013, at 9:30 PM, Jingkun Gao wrote:

What's the difference between test error and train error?

I thought there is only one kind of error. For example, in N-fold cv, each cv will generate an error, which is calculated using wrong labelled test samples divide total test samples. We just sum up the errors in the loop and divided by N. What's the name of this error? 

Thanks,

Regards,

Jingkun

--
http://alex.smola.org/teaching/cmu2013-10-701 (course website)
http://www.youtube.com/playlist?list=PLZSO_6-bSqHQmMKwWVvYwKreGu4b4kMU9 (YouTube playlist)
---
You received this message because you are subscribed to the Google Groups "10-701 Spring 2013 CMU" group.

To unsubscribe from this group and stop receiving emails from it, send an email to 10-701-spring-2013-cmu+unsub...@googlegroups.com.