This topic contains 4 replies, has 2 voices, and was last updated by Federico Pascual 1 year, 9 months ago.
Is there other ways to understand how my classifier works besides confusion matrix?
Hey guys! One quick question: is there other ways to understand how my classifier works besides confusion matrix? Cheers!
Hey Colin! That’s a great question :)
Besides the confusion matrix, we recommend to use the classifier stats to understand how your classifier works and to about potential opportunities of improvement.
The accuracy is the most basic performance measure, for the selected non-leaf category it shows the percentage of test texts that where classified to the correct subcategory in the evaluation process. Since leaf categories doesn’t have children this value doesn’t make sense and won’t be displayed.
The precision for a non-root category is the percentage of the test samples that were classified to this category by it’s parent and actually belonged to this category. Since the root has no parent this value doesn’t make sense and won’t be displayed.
The recall for a category is the percentage of all the test samples that originally belonged to this category and in the evaluation process were correctly classified to this category by it’s parent. Since the root has no parent this value doesn’t make sense and won’t be shown.
Using all stats
Is key to understand that you should take a look at all stats to know how your classifier is performing. Depending on your particular application, you may be more interested in having better precision or recall (usually there’s a trade off between them).
Hope this helps!
Ryan April September 15, 2015 at 2:24 pm
- This reply was modified 1 year, 9 months ago by Raul Garreta.
You forgot to mention debug out, when testing a train classifier.Ryan April September 15, 2015 at 2:25 pm
I meant debug output..Federico Pascual September 15, 2015 at 3:11 pm
That’s right Ryan! Debug output is also a great way to understand how a classifier works. Thanks for pointing it out :)