0
$\begingroup$

Is there an easy way to calculate for each train+test set the chance level accuracy? my models deal with different data sets that have very large number of classes and they are not uniform distributed.

$\endgroup$
0
$\begingroup$

So i figured it out, the way to find the chance level accuracy for multiclass data set with imbalanced classes we can use scikit learn dummy calssifier with the following configuration :

DummyClassifier(strategy='stratified')

“stratified”: generates predictions by respecting the training set’s class distribution." DummyClassifier on scikit documentation

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.