Re: st: Intervals for predicted mean value vs predicted individual value

[Jorge Eduardo Pérez Pérez <jorge_perez@brown.edu>]

You just need to add a new observation to perform an out of sample
prediction, then use -predict- as you have been doing.

clear
input temp fuelcons
28 12.4
28 11.7
32.5 12.4
39 10.8
45.9 9.4
57.8 9.5
58.1 8
62.5 7.5
end
* Perform regression
reg fuelcons temp
* Add one observation
set obs `=_N+1'
* Set value of temp for out of sample prediction
replace temp=40 in l/l
* Calculate prediction and standard erros of mean and forecast
predict pfuelcons
predict pmfuelcons, stdp
predict pyfuelcons, stdf
* Calculate cis
* Set Degrees of freedom and confidence level
glo df=e(df_r)
glo conf=0.05
* Calculate ci in new variables.
gen ciml=pfuelcons - invttail(${df},${conf}/2)*pmfuelcons
gen cimh=pfuelcons + invttail(${df},${conf}/2)*pmfuelcons
gen cipl=pfuelcons - invttail(${df},${conf}/2)*pyfuelcons
gen ciph=pfuelcons + invttail(${df},${conf}/2)*pyfuelcons
* Display
li pfuelcons ci* in l/l

I don't know if this can be done in a single command.


--------------------------------------------
Jorge Eduardo Pérez Pérez
Graduate Student
Department of Economics
Brown University


On Mon, Feb 24, 2014 at 1:32 AM, Adam Guerrero <adamgue@gmail.com> wrote:
> Dear Statalist,
>
> I am trying to reproduce an example in Bowerman et al.'s "Forecasting,
> Time Series, and Regression."
>
> I have inputted the following data into Stata
>
> . input temp fuelcons
> . 28 12.4
> . 28 11.7
> . 32.5 12.4
> . 39 10.8
> . 45.9 9.4
> . 57.8 9.5
> . 58.1 8
> . 62.5 7.5
> . end
>
> After regressing fuelcons on temp, I can obtain predicted values of
> fuelcons using
>
> . regress fuelcons temp
>
> . predict pfuelcons
>
> I can also get the standard errors of the predicted mean using
>
> . predict pmfuelcons, stdp
>
> Finally, I can get the standard errors of the predicted individuals
> fuelcons values using
>
> . predict pyfuelcons, stdf
>
> I know that I can use this information to obtain a confidence interval
> for the conditional mean and a confidence interval for a particular
> value of fuelcons given a value for temp in the data.
>
> My question is whether there is a command that will allow me to
> estimate a 95% confidence interval for the mean value of fuelcons
> given a particular value of temp that is not included in the data, but
> is in the range of temp (e.g., given a temp of 40).
>
> I would also like to get a 95% confidence interval for an individual
> value of fuelcons given a particular value of temp that is not
> included in the original data, but is in the range of temp (e.g.,
> given a temp of 40).
>
> The confidence interval for the mean value of fuelcons when temp
> equals 40, and the prediction interval for an individual value of
> fuelcons when temp equals 40 are computed automatically in MINITAB as
>
> Fit            SE Fit                95% CI                          95% PI
> 10.721     .241                  (10.130, 11.312)          (9.015, 12.427)
>
> Is it possible to do this in Stata with a single command?
>
> I am a bit worried that I am overlooking something simple in, so I
> apologize in advance if that's the case, and thank you for assisting
> with such a long question.
>
> Cheers,
> Adam Guerrero
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/