1
$\begingroup$

We know that weather conditions in China is not very useful to predict price of a house in Spain (as life experience). However, when we drop that China's weather condition, the accuracy is reduced largely. Will we keep it?

$\endgroup$
2
  • 1
    $\begingroup$ Did the accuracy drop in the out of sample test set as well? I did be surprised if so. $\endgroup$ Commented May 8, 2021 at 14:32
  • $\begingroup$ hmm i think you got the point. Thanks $\endgroup$
    – Lucifer
    Commented May 8, 2021 at 14:34

1 Answer 1

1
$\begingroup$

It would mean that the model trained with some irrelevant features is overfit: a good model uses only features which have some predictive power on the target variable, so if the sample is representative enough any irrelevant feature is ignored or assigned very little weight. However if the training dataset is too small or the model too complex then it uses details which happen by chance, so it could use some irrelevant features: this is overfitting.

In case the performance on the test set was high despite the model being overfit, it's likely that there is some data leakage additionally.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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