O'Reilly logo
  • Richard Xie thinks this is interesting:

scores = cross_val_score(clf1, X, y, cv=3,scoring='accuracy', n_jobs=-1)

From

Cover of Python: Real World Machine Learning

Note

This is wrong. You are using the testing data to do the cross-validation. You should have used
accuracy = accuracy_score(y, pred2)
to compute the prediction accuracy.
Then the true accuracy is around 0.83, not the claimed 0.92.