Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: smoothing quantiles
From
Nick Cox <[email protected]>
To
[email protected]
Subject
Re: st: smoothing quantiles
Date
Sat, 2 Apr 2011 18:23:39 +0100
I did not say that quantile smoothing is harder than density
estimation, and I don't think it is.
What you want sounds more like quantile regression to me. See -qreg-.
Nick
2011/4/2 László Sándor <[email protected]>:
> Thank you, Nick, this is very useful.
>
> It turns out that I myself would have been more interested in
> something like density estimation (well, more like CDFs), but instead
> of 3D plots of density (I know that this is possible), using "level
> sets" or contour curves in 2D, which basically plotting the quantiles
> of yvar in terms of values (not quantiles) of xvar.
>
> The code you just sent does not seem to do that. But maybe this is
> only how we can plot the results of hdquantile? I should play with it
> more. In any case, you say that CDFs (quantiles) are harder to smooth
> than densities? This makes sense. But for descriptive purposes I would
> still try to do that in a dataset rich (large) enough.
>
> Am I wrong on this? If not, do you know of an easyish workaround?
> On Sat, Apr 2, 2011 at 3:13 AM, Nick Cox <[email protected]> wrote:
>> One way of seeing what -hdquantile- does is, naturally, graphical.
>>
>> sysuse auto
>> hdquantile mpg, gen(hdq_mpg)
>> qplot *mpg , c(. l) ms(Oh none)
>>
>> where -qplot- is from SJ.
>>
>> It seems curious to me that quantile smoothing is far less practised
>> than density estimation.
>>
>> Nick
>>
>> On Fri, Apr 1, 2011 at 5:50 PM, Nick cox <[email protected]> wrote:
>>
>>> I implemented one method in -hdquantile- from SSC.
>>
>>> On 1 Apr 2011, at 17:43, László Sándor <[email protected]> wrote:
>>>
>>>> Hi all,
>>>>
>>>> I would like to plot smooth predicted quantiles with Stata 11.2 on
>>>> Unix (console).
>>>>
>>>> My goal is to have something like what lowess is for the mean for the
>>>> yvar. Preferably, it would be nice to be able to plot multiple
>>>> quantiles in the same plot (e.g., all 3 quartiles). This would be a
>>>> useful way to summarize the distribution of a yvar across the values
>>>> of an xvar.
>>>>
>>>> I'd rather not reinvent the wheel on this one --- I hope there is
>>>> something like this already out there (or built in), only I could not
>>>> find it.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/