0
$\begingroup$

I want to plot C% confidence intervals for a regressor (GradientBoosting in my case).

I found this example on scikit-learn documentation https://scikit-learn.org/stable/auto_examples/ensemble/plot_gradient_boosting_quantile.html#sphx-glr-auto-examples-ensemble-plot-gradient-boosting-quantile-py where the area between the quantiles (alpha) 0.05 and 0.95 is indicated as 95% confidence interval.

Two questions:

  • Is correct the scikit-learn example? Shouldn't be 90% confidence interval with those quantiles equal to 0.05 and 0.95?
  • If I want to plot confidence interval of C%, is it correct to pick alpha_low = (1-C)/2 and alpha_high = (1 - (1-C)/2) ?

thank you!

Improve this question
$\endgroup$
3
  • $\begingroup$ Is the quantile-regression tag relevant here? $\endgroup$
    – Richard Hardy
    Commented Nov 19, 2020 at 14:04
  • $\begingroup$ a regression problem with 'quantile' loss is related to quantile-regression, isn't it? $\endgroup$
    – user1403546
    Commented Nov 19, 2020 at 14:33
  • $\begingroup$ This is not obvious from your post, though maybe it is hidden under the link. In that case it is OK. $\endgroup$
    – Richard Hardy
    Commented Nov 19, 2020 at 14:54

0

Reset to default

Your Answer

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