In the process you did, you just evaluated the model that you already built
using some instances (external test set). You did not build the main model
itself here. Thus, there is no overfitting can be occurred at this stage.
However, at early stage we generally obtain large number of data to create
genral prediction model and we wish that it's a good model. After that, we
can deploy this model and use it on an external test set file.
On 14 Nov 2015 21:02, "Nada Courses" <snada.g14(a)gmail.com> wrote:
I ran some classifiers on small data size (30) instances , and I got 90%
accuracy on training ,and 100% accuracy on test & ROC=1!
How I know that the model is good and there is no over fitting ?
training =22 instances ,and test=8 instances .
Wekalist mailing list
Send posts to: [hidden email]
List info and subscription status:
If you reply to this email, your message will be added to the discussion
To unsubscribe from WEKA, click here