I have been given with a set of data, which is supposedly come from an unknown distribution F. And I am asked to propose a suitable parametric or nonmparametric estimator for quantiles q(α) =F⁻¹(α) for the series of α. From plotting the data as a histogram, I see that the data has a positively skewed distribution, but I`m stuck from there.

  • $\begingroup$ Is this a homework assignment? $\endgroup$ – jbowman May 3 '19 at 15:54

The simplest estimator would simply be the empirical quantiles from your dataset. Alternatively, you could feed your data into a kernel density and take quantiles from that. Or finally, you could fit a known distribution (if possible, guided by substantive knowledge of your data - if you know it's nonnegative, then there are likely better alternatives than the normal distribution).

If your data depend on covariates, you could try analogues: quantile regression, functional data analysis, or an appropriate regression model.

| cite | improve this answer | |
  • $\begingroup$ Thank you for the advice! I will try with the empirical quantile and see if it works. $\endgroup$ – wruskrappy May 3 '19 at 22:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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