CODEWITHSUNDEEP
scikit-learncross-validation
Ricardo Cruzado Valladares
Ricardo Cruzado Valladares

Reputation: 19

cross_val_score in Ridge/Lasso regression

I am running a Ridge regression in a Jupyter notebook, which I attach here.

ridge = Ridge(alpha=optimal_ridge.alpha_)

ridge_scores = cross_val_score(ridge, Xs, y, cv=10)

print(ridge_scores)
print(np.mean(ridge_scores))`

When applying the funcion cross_val_score to my estimator, what metric does it give me? I obtain 10 values lower than 1 (that come from the cross validation), which are:

[0.90397003 0.88391849 0.77181566 0.85037008 0.85705655 0.88257961
 0.9114154  0.91447498 0.93543431 0.93167352]

Thank you!

Upvotes: 1

Views: 1591

Answers (1)

Alperen
Alperen

Reputation: 4602

According to the docs, it uses "the estimator’s default scorer (if available)". In your case, the estimator is Ridge and its score function returns "the coefficient of determination R^2 of the prediction" as stated in its docs.

Also, you can define your own scoring function and pass it to cross_val_score as a parameter (docs):

def your_scorer(estimator, X, y):
    ...

cross_val_score(ridge, Xs, y, cv=10, scoring=your_scorer)

Upvotes: 1

Related Questions