Re: st: How to get parts of XB matrix

[Austin Nichols <austinnichols@gmail.com>]

Zhi Su <su.zh@husky.neu.edu>:
If you want a prediction zeroing out only a few coefficients, it's
probably easier to start with -predict- and then subtract those few
off.  But you should get the same result either way.

sysuse auto, clear
reg  price mpg trunk weight length turn
predict p
g pr=p-_b[_cons]
g xb=0
loc X mpg trunk weight length turn
foreach v of loc X {
 replace xb=xb+`v'*_b[`v']
 }
su pr xb


On Fri, Jun 3, 2011 at 5:56 PM, Austin Nichols <austinnichols@gmail.com> wrote:
> Zhi Su <su.zh@husky.neu.edu> :
> Something like
> g xb=0
> foreach v of loc X {
> replace xb=xb+'v'*_b['v']
> }
>
> On Friday, June 3, 2011, Zhi Su <su.zh@husky.neu.edu> wrote:
>> Dear statalist,
>>
>>   I run a regression Y=XB+u, with 30 independent variables. I want to
>> get predicted XB of 15 independent variables among 30 independent
>> variables.
>>   Here are my questions
>>   1.The coefficient of an independent variable x is stored in _b[x].
>> Is _b[x] a vector or scalar?
>>   2. Generally, I use "local" to assign X I want in to a local macro
>> name. And I can easily change X in the local macro name. How can I
>> utilize this local macro name when I create XB matrix with independent
>> variables I want?

*
*   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/