In quantile regression (https://en.wikipedia.org/wiki/Quantile_regression), what are some suitable evaluation metrics?
Intuitively, I think a good model should have:
- good accuracy, i.e. the prediction interval (say at 90% confidence) should contain the true value of
y
. - narrow width, i.e. the interval should not be large (if we expand the interval to
[-Inf, Inf]
definitely this interval will cover the true ofy
).
Could you recommend some ways to evaluate a predictor that outputs the prediction interval?
Many thanks