0
$\begingroup$

I am using catboost for a multiclass classification problem. I want to use quadratic weighted kappa as the evaluation metric.

Catboost already has WKappa as an eval_metric but it is linearly weighted variant, not quadratic.

clf = CatBoostClassifier(objective='MultiClass',
eval_metric=q_weighted_kappa(),
learning_rate=0.01,
n_estimators=750,
od_type="Iter",
depth=35,
early_stopping_rounds=500,
random_seed=42,

)

I need help to define the function q_weighted_kappa() that I can use as specified above.

$\endgroup$
0
$\begingroup$

Sklearn kappa has 2 variants: simple and weigthed. It seems WKappa refers to weighted (quadratic) kappa in terms of sklearn, and Kappa - to simple.

$\endgroup$

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.