Renaming rows in Matrix

[Zuhumnan Dapel]

Dear All,
I just obtained the following matrix

X[18,4]
         imeu0      imeu1      imeu2      imeu3
 r1  .17012477  .14678881  .14600933  .30892292
 r2  .16162819  .13678399  .14205745  .27877864
 r3  .17284968   .1417264  .14267236  .30505028
 r4  .15048751  .13105537  .13583121  .29497504
 r5  .16008531  .13823907  .14575586  .31288761
 r6  .16042244  .13621108  .15766013  .32482761
 r7  .17476352  .14296345  .15441483  .33454451
 r8  .19413681  .14554305  .15211919  .33655742
 r9  .21324582  .15600634  .15217966  .34234405
r10  .22980724  .17459309  .16466963  .39787501
r11  .24465741  .16848315  .17576084  .43677589
r12  .27512023  .18187349  .18600267  .46879619
r13  .28697997  .19372031  .20313573  .50782186
r14  .29941869  .20602649   .2135499  .54286134
r15   .2944608  .21142302   .2167595  .54684091
r16    .363224  .21042196  .21408814  .56670946
r17  .37690601  .22233553  .21801062  .62060088
r18  .37223187  .23511408  .23473491  .57701015

How do I rename the rows r1, r2, ..., r18 to 1990, 1991,..., 2007?

Thanks,
Dapel

[Clyde Schechter]

local rownames
forvalues j = 1/18 {
    local rownames `rownames' `=1989+`j''
}
matrix rownames X = `rownames'
matrix list X

This will work, at least this far. I have never before, however, seen a Stata matrix whose row or column names begin with digits. A quick review in the user manuals has not turned up anything to say that it is illegal, but it is also clear from the text and all examples that Stata matrix row and column names are often intended to match variable names--and, of course, a variable name cannot begin with a digit. I don't know what you plan to do with this matrix once you do this. It may well be that some subsequent matrix manipulations will fail, perhaps in strange ways, with these names. If it doesn't otherwise defeat your purpose, I would play it safe and rename the rows to yr1990, yr1991, etc. instead using the obvious modification of my code.

[Zuhumnan Dapel]

Absolutely. Thanks!

Dapel

[Rich Goldstein]

re: Clyde's post (#2 above), I have used matrices with rownames starting with digits - no problem

[Zuhumnan Dapel]

Good to know this too. Thanks Rich

[Nick Cox]

No need for a loop here.

 
. mat z = J(18,4,42)

. numlist "1990/2007"

. mat rownames z = `r(numlist)'

. mat li z

z[18,4]
      c1  c2  c3  c4
1990  42  42  42  42
1991  42  42  42  42
1992  42  42  42  42
1993  42  42  42  42
1994  42  42  42  42
1995  42  42  42  42
1996  42  42  42  42
1997  42  42  42  42
1998  42  42  42  42
1999  42  42  42  42
2000  42  42  42  42
2001  42  42  42  42
2002  42  42  42  42
2003  42  42  42  42
2004  42  42  42  42
2005  42  42  42  42
2006  42  42  42  42
2007  42  42  42  42

[Zuhumnan Dapel]

Quite a straight one Dr Cox. Thanks.

[Sean Fiedler]

Is there a way to get rownames that would be something like 1990q3 to 1995q4?

[Sean Fiedler]

I found a not quite as nice solution to my question:

gen tstr = string(t, "%tq")
levelsof tstr if inrange(t, t[10], t[15]), local(myrownames)     // choose whatever your desired time periods are
matrix rownames X = `myrownames'

[Clyde Schechter]

It's a bit cumbersome, but:

mat z = J(22, 4, 42)
local rn
forvalues i= `=tq(1990q3)'/`=tq(1995q4)' {
    local rn `rn' `:display %tq `i''
}
mat rownames z = `rn'
matrix list z

[Sean Fiedler]

Thank you Clyde!

[Nick Cox]

You can get a set of names in a local by pushing the dates in and out of Mata:

. mata
: t = (yq(1990,3)..yq(1995,4))

: t
         1     2     3     4     5     6     7     8     9    10    11    12
    +-------------------------------------------------------------------------
  1 |  122   123   124   125   126   127   128   129   130   131   132   133
    +-------------------------------------------------------------------------
        13    14    15    16    17    18    19    20    21    22
     -------------------------------------------------------------+
  1    134   135   136   137   138   139   140   141   142   143  |
     -------------------------------------------------------------+

: names = strofreal(t, "%tq")

: names
            1        2        3        4        5        6        7        8
    +-------------------------------------------------------------------------
  1 |  1990q3   1990q4   1991q1   1991q2   1991q3   1991q4   1992q1   1992q2
    +-------------------------------------------------------------------------
            9       10       11       12       13       14       15       16
     -------------------------------------------------------------------------
  1    1992q3   1992q4   1993q1   1993q2   1993q3   1993q4   1994q1   1994q2
     -------------------------------------------------------------------------
           17       18       19       20       21       22
     -------------------------------------------------------+
  1    1994q3   1994q4   1995q1   1995q2   1995q3   1995q4  |
     -------------------------------------------------------+

: end

So,

 
. mata : st_local("names", invtokens(strofreal((yq(1990, 3)..yq(1995,4)), "%tq")))

. di "`names'"
1990q3 1990q4 1991q1 1991q2 1991q3 1991q4 1992q1 1992q2 1992q3 1992q4 1993q1 1993q2 1993q3 1993q4 1994q1 1994q2 1994q3 1994q4 1995q1 1995q2 1995q3 1995q4

.

Hence in your case you would just use the macro for your matrix row names.

[Sean Fiedler]

Thank you Nick, appreciate it