0
$\begingroup$

I have made a learning curve that looks like this:

enter image description here

Why wouldn't it be more like both training and cross-validation score begin low and both gradually increase with more samples? Why does one start high while the other starts low? For example, this is how I thought it would look: enter image description here

Thanks

Improve this question
$\endgroup$
4
  • $\begingroup$ I cannot see any images $\endgroup$
    – cbeleites
    Commented Mar 10, 2015 at 14:51
  • $\begingroup$ What is "score"? What kind of metric do you use? $\endgroup$
    – Artem Sobolev
    Commented Mar 11, 2015 at 23:09
  • $\begingroup$ take a look at this video to understand learning curves : youtube.com/watch?v=ISBGFY-gBug $\endgroup$
    – DVG
    Commented Apr 15, 2017 at 20:41
  • $\begingroup$ See also scikit-learn.org/stable/modules/…, since this is tagged with scikit-learn $\endgroup$
    – steffen
    Commented Mar 12, 2018 at 13:43

1 Answer 1

Reset to default
1
$\begingroup$

Because a sufficiently powerful machine learning algorithm (together with nicely separable data), will have no problem correctly classifying a small number of data points. Of course that usually means it will be overfitting to those few points, which is why your training score is high and your cross-validation score low.

Improve this answer
$\endgroup$

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.