Adding variable names to name column of matrix

[Fabian Fuchs]

Dear Statalist-users,

I would like to save the names of all companies c_* to use them as rownames of matrix mat1. Which works now that I created locals for every year `l'. I don´t know why dropping `rownnames' creats a comformatability error, looping through the years l.

For the matrix beta I would like to have the rownames of the companies and then the year. The problem is that the year I am currently in is just added to the very last company each year. I tried to use strings or generate a matrix or varialbe to store the rownames, but nothing works.

forvalues l = 1995(1)2015 {
...

foreach companies of varlist c_* {

...


local rownames_`l' `rownnames_`l'' `companies'

matrix rownames mat1 = `rownames_`l''

matrix rownames beta_`l' = "`rownames_`l''_`l'"

Thank you very much

[Lance Erickson]

Fabian,

It's really, really hard to tell what's working and what's now working or why from your explanation and code. Knowing a bit more about the data structure might help too. Do you want the matrix to have row names like...

Stata1995
Stata1996
.
.
.
Stata2015
SAS1995
SAS1996
.
.
.
SAS2015

I can imagine a data structure where it wouldn't be too hard to get that but I have no idea if that's the data structure you're working with. It's hard to help you move forward without a better sense for your data and what exactly is going wrong and where.

Lance

[Fabian Fuchs]

Hi Lance,

I have variables named c_1, c_2, ... for which I run regressions. For the years "l" from 1995 until 2015 I get beta values.

Now I want to store those values, which works fine. But I have troubles naming the rows of my matrix correctly.

What I want is:

c_1_1995
c_2_1995
...
c_1_1996
...

What the code does:

c_1
...
c_last_1995
c_2
...
c_last_1996

 matrix rownames beta_`l' = "`rownames_`l''_`l'"

If I just use

 matrix rownames beta_`l' = `rownames_`l''_`l'

and get rid of the "" - I get.

c_1 c_2 c_..._1995
c_1 c_2 c_..._1995
...
c_1 c_2 c_..._1996
...






The first part of the question was: Why

 local rownames `rownnames' `companies'  local drop rownames

won´t work. Stata gives back a conformatability error each time. I thought the local rownames stores the variable names of companies (c_1 c_2 ... ) in rownames and after dropping rownames (because the amount of companies in the year-sample changes) can be used again.

By adding _`l' it worked, but I don´t understand why different locals were needed to do the job.

Thank you

[Fabian Fuchs]

I tried to get a solution with another local for the years

  
local rowname_`l' `rowname_`l'' `companies'
local year `year'  `l'

matrix rownames beta_`l' = "`rowname_`l''_`year'"

what I get now is

c_1 c_2 .... 1995 1995 1995 ....
c_1 c_2 .... 1995 1995 1995 ....
c_1 c_2 .... 1995 1995 1995 ....
...
c_1 c_2 .... 1996 1996 1996 ....
c_1 c_2 .... 1996 1996 1996 ....

The problem seems to be that the local rowname looks like this

c_1 c_2 ...

and year

1995 1995 ...

putting only one of the two in as rownames of the matrix beta_1995, Stata recognizes to use it like

c_1
c_2

or

1995
1995

but combining those two, looks like the this c_1 c_2 .... 1995 1995 1995 ....

Thanks

[Lance Erickson]

How about sharing a snippet of data (see dataex on SSC)? I'm afraid I won't have any other ideas without seeing exactly what you're working with.