Quantile Regression for Panel Data. R square

[Zara Ruzim]

Good day everyone,
I was kindly requesting assistance. I applied Quantile Regression for Panel Data and everything works. But I cannot find R squared. My question how can see R squared.
I've tried different methods that I've seen on this forum but I think I am not doing something.
I will appreciate it if someone corrects my code to me.
My code:
qregpd yield_msavi avg8_msavi avg8_cired avg8_ndre , id(fid_msavi) fix(year_msavi) optimize(mcmc) noisy draws(1000) burn(100) arate(.75)

[FernandoRios]

Hi Zara
It most likely that qregpd does not have an Rsquare to begin with. (or the R2 has no meaning in this setting, so they skipped it.
HTH
F

[Zara Ruzim]

Dear FernandoRios,

Thank you for your reply. But I need to see R2 square.

Best regards,

[Nick Cox]

https://www.stata.com/support/faqs/s...ics/r-squared/ may help. I suspect that many statistical people would regard that FAQ as too liberal even taken as sceptically.as it was written. (I am well acquainted with the views of the FAQ author, being the FAQ author.)

In short, you need to get predictions out of your model and if you can do that, then you could correlate observed and predicted outcomes. But that has little or no bearing on the merits of the model and zero bearing on how the model is defined or its parameters estimated in this case.

The classical view of regression includes asking how far mean outcomes are predictable. Even if you focus on the 0.5 quantile or median, that is not necessarily how best to think about quantile regression.

When software authors don't provide R-square, the question is always Why not? and it's best to assume a good reason.

[Zara Ruzim]

Dear Nick Cox,

Thank you for your reply.

I have one independent variable and four dependent variables. All my variables in the panel data format.
I would like to use quantile regression for panel data and look at how well it works in predicting that small difference between a so-so harvest and a harvest I would consider as default.
I would much appreciate it if you advise me. I would get a final result like my the screen. Pls, find below the screen.

Sorry for some mistakes.

Best regards,

[Nick Cox]

I think I understand what you're asking for (although you have one dependent variable and four independent variables); I don't have specific suggestions for how to get such a number out of qregpd which I have never used. Also, it is quite possible that there are extra specific reasons why the idea makes little or no sense for that model: I haven't read the literature either.

[Zara Ruzim]

Ok. Thank you for your reply.