Skip to main content
Cross Validated

Return to Question

Notice removed Draw attention by CommunityBot
Bounty Ended with David Dale's answer chosen by CommunityBot
Tweeted twitter.com/StackStats/status/928971179863535616
Notice added Draw attention by Clément F
Bounty Started worth 50 reputation by Clément F
added 55 characters in body
Source Link
Statlearner
Statlearner
  • 161
  • 2
  • 7

While trying to perform some quantile regressions on some data I have encountered a classic problem: the 0.25,0.50, 0.75, 0.95 quantile curves are crossing each other (I am dealing with linear curves).

I am currently working with Stata using the sqreg command.

I have read Howard Bondell's paper "Non-crossing quantile regression curve estimation"(2010) and I know he used R to code a solution.

I have also read papers from Andriyana and from Chernozhukov (on the issue of non-monotonicity issues in quantile regressions).

Just to be clear about what we are talking, here below is an example of quantile curves that DO NOT CROSS (and my issue is what to do when they cross):

image

I was wondering if any of you could recommend what software solutions are most commonly used in the working practice, to deal with crossing quantile curves (=to avoid their crossing), I know there are various methods, so the question is both: what method you prefer and what software would you to use to solve the issue?

Note: I am not posting a sample of my data because it's irrelevant, I am just interested in the general theory of the problem and how to deal with it specifically with software (ideally STATA or Mathematica, or R or Python or Perl).

While trying to perform some quantile regressions on some data I have encountered a classic problem: the 0.25,0.50, 0.75, 0.95 quantile curves are crossing each other (I am dealing with linear curves).

I am currently working with Stata using the sqreg command.

I have read Howard Bondell's paper "Non-crossing quantile regression curve estimation"(2010) and I know he used R to code a solution.

I have also read papers from Andriyana and from Chernozhukov (on the issue of non-monotonicity issues in quantile regressions).

Just to be clear about what we are talking, here below is an example of quantile curves that DO NOT CROSS (and my issue is what to do when they cross):

image

I was wondering if any of you could recommend what software solutions are most commonly used in the working practice, to deal with crossing quantile curves (=to avoid their crossing), I know there are various methods, so the question is both: what method you prefer and what software would you to use to solve the issue?

Note: I am not posting a sample of my data because it's irrelevant, I am just interested in the general theory of the problem and how to deal with it specifically with software.

While trying to perform some quantile regressions on some data I have encountered a classic problem: the 0.25,0.50, 0.75, 0.95 quantile curves are crossing each other (I am dealing with linear curves).

I am currently working with Stata using the sqreg command.

I have read Howard Bondell's paper "Non-crossing quantile regression curve estimation"(2010) and I know he used R to code a solution.

I have also read papers from Andriyana and from Chernozhukov (on the issue of non-monotonicity issues in quantile regressions).

Just to be clear about what we are talking, here below is an example of quantile curves that DO NOT CROSS (and my issue is what to do when they cross):

image

I was wondering if any of you could recommend what software solutions are most commonly used in the working practice, to deal with crossing quantile curves (=to avoid their crossing), I know there are various methods, so the question is both: what method you prefer and what software would you to use to solve the issue?

Note: I am not posting a sample of my data because it's irrelevant, I am just interested in the general theory of the problem and how to deal with it specifically with software (ideally STATA or Mathematica, or R or Python or Perl).

Source Link
Statlearner
Statlearner
  • 161
  • 2
  • 7

The issue of quantile curves crossing each other

While trying to perform some quantile regressions on some data I have encountered a classic problem: the 0.25,0.50, 0.75, 0.95 quantile curves are crossing each other (I am dealing with linear curves).

I am currently working with Stata using the sqreg command.

I have read Howard Bondell's paper "Non-crossing quantile regression curve estimation"(2010) and I know he used R to code a solution.

I have also read papers from Andriyana and from Chernozhukov (on the issue of non-monotonicity issues in quantile regressions).

Just to be clear about what we are talking, here below is an example of quantile curves that DO NOT CROSS (and my issue is what to do when they cross):

image

I was wondering if any of you could recommend what software solutions are most commonly used in the working practice, to deal with crossing quantile curves (=to avoid their crossing), I know there are various methods, so the question is both: what method you prefer and what software would you to use to solve the issue?

Note: I am not posting a sample of my data because it's irrelevant, I am just interested in the general theory of the problem and how to deal with it specifically with software.