How to rename around 200 rows of a matrix simultaneously or in iteration

[Siying Li]

Hi, I'm very new to Stata and I've searched and tried answers for this question for three hours. I'm using Stata16. I have covid related data (daily cumulative/new cases/deaths data) of each country and I would like to set row's name to be according country's name. To examine on each country's data, I firstly split the data set to 195 dta files:
levelsof location, local(level1)
foreach x of local level1 {
preserve
keep if location=="`x'"
save file`x',replace
restore
}

Then I repeat a regression on each country's data using foreach, resulting in a matrix of R^2

I've renamed columns. However, I don't know how to rename around 200 rows easily.

I have two ideas. I firstly tried to use levelsof to extract location names from original covid related dataset and set it to be the rows' names:
frame owid: levelsof location, local(rownames)
matrix rownames R2 = `rownames'
However, it seems that the dataset is another frame, the list of locations cannot be row names? The row names were not changed.

Then, I tried to rename each row in iteration:
local myfilelist: dir . files "*.dta"
local rownames
foreach filename of local myfilelist {
use `filename', clear
local rownames `rownames' `filename'
}
matrix rownames R2 = `rownames'
But I got error message: fileafghanistan: operator invalid

I'm sorry if I am not clear in questioning. I previously used other softwares and Stata is quite new for me. I will appreciate if you can give some advice on codes.

Thank you for your time and help!

[Andrew Musau]

You are better off using ISO country codes to identify countries. Country names violate Stata's naming conventions, most prominently, no spaces. Otherwise, see

help strtoname()

that may help with converting strings with invalid characters to names. The bigger problem with your example is that it appears you do not have control over what is stored in your local "rownames". With valid names, -levelsof- can create the local for you, and it sorts the names alphabetically. So when creating the matrix, make sure that the rows correspond to an alphabetical sort order. Here is an example:

sysuse census, clear
*STATE2 IS THE 2 LETTER STATE IDENTIFIER
sort state2
mkmat pop popurban, mat(P)
levelsof state2, local(states) clean
mat rownames P= `states'
mat l P

Res.:

. mat l P

P[50,2]
         pop  popurban
AK    401851    258567
AL   3893888   2337713
AR   2286435   1179556
AZ   2718215   2278728
CA  23667902  21607606
CO   2889964   2329869
CT   3107576   2449774
DE    594338    419819
FL   9746324   8212385
GA   5463105   3409081
HI    964691    834592
IA   2913808   1708232
ID    943935    509702
IL  11426518   9518039
IN   5490224   3525298
KS   2363679   1575899
KY   3660777   1862183
LA   4205900   2887309
MA   5737037   4808339
MD   4216975   3386555
ME   1124660    534072
MI   9262078   6551551
MN   4075970   2725202
MO   4916686   3349588
MS   2520638   1192805
MT    786690    416402
NC   5881766   2822852
ND    652717    318310
NE   1569825    987859
NH    920610    480325
NJ   7364823   6557377
NM   1302894    939963
NV    800493    682947
NY  17558072  14858068
OH  10797630   7918259
OK   3025290   2035082
OR   2633105   1788354
PA  11863895   8220851
RI    947154    824004
SC   3121820   1689253
SD    690768    320777
TN   4591120   2773573
TX  14229191  11333017
UT   1461037   1233060
VA   5346818   3529423
VT    511456    172735
WA   4132156   3037014
WI   4705767   3020732
WV   1949644    705319
WY    469557    294639

[Siying Li]

You are better off using ISO country codes to identify countries. Country names violate Stata's naming conventions, most prominently, no spaces. Otherwise, see

help strtoname()

that may help with converting strings with invalid characters to names. The bigger problem with your example is that it appears you do not have control over what is stored in your local "rownames". With valid names, -levelsof- can create the local for you, and it sorts the names alphabetically. So when creating the matrix, make sure that the rows correspond to an alphabetical sort order. Here is an example:

sysuse census, clear
*STATE2 IS THE 2 LETTER STATE IDENTIFIER
sort state2
mkmat pop popurban, mat(P)
levelsof state2, local(states) clean
mat rownames P= `states'
mat l P

Res.:

. mat l P

P[50,2]
pop popurban
AK 401851 258567
AL 3893888 2337713
AR 2286435 1179556
AZ 2718215 2278728
CA 23667902 21607606
CO 2889964 2329869
CT 3107576 2449774
DE 594338 419819
FL 9746324 8212385
GA 5463105 3409081
HI 964691 834592
IA 2913808 1708232
ID 943935 509702
IL 11426518 9518039
IN 5490224 3525298
KS 2363679 1575899
KY 3660777 1862183
LA 4205900 2887309
MA 5737037 4808339
MD 4216975 3386555
ME 1124660 534072
MI 9262078 6551551
MN 4075970 2725202
MO 4916686 3349588
MS 2520638 1192805
MT 786690 416402
NC 5881766 2822852
ND 652717 318310
NE 1569825 987859
NH 920610 480325
NJ 7364823 6557377
NM 1302894 939963
NV 800493 682947
NY 17558072 14858068
OH 10797630 7918259
OK 3025290 2035082
OR 2633105 1788354
PA 11863895 8220851
RI 947154 824004
SC 3121820 1689253
SD 690768 320777
TN 4591120 2773573
TX 14229191 11333017
UT 1461037 1233060
VA 5346818 3529423
VT 511456 172735
WA 4132156 3037014
WI 4705767 3020732
WV 1949644 705319
WY 469557 294639

Hi Andrew,

Many thanks for your reply. I will try to use ISO country codes instead of country names. I noticed that you mentioned -levelsof- sorts alphabetically in reply for someone else, so I ensured it can work for my dataset. I wonder what should I do to control local rownames? I am quite confused about this part. As I understand, all R-squared are calculated using dta files in a directory, and aimed rownames are extracted from another dta file named owid. Can local rownames defined in frame owid be used in R2 matrix constructed using other dta files? I'm quite struggled with separate frames in Stata. I think variable m defined in frame A cannot be used in operations on frame B, right? Sorry, I'm too confused about the processing.

[Andrew Musau]

I do not think that's it. This works:

clear
mat A= 1\2\3\4
mat l A
frame create new
frame new{
    sysuse auto, clear
    gen make2= strtoname(make)
    levelsof make2 if _n<5, local(names) clean
}
mat rownames A= `names'
frame drop new
mat list A

As I stated, you probably need to pass the strings through -strtoname()- before running -levelsof-. If this does not help, post a data example using the dataex command of the variable containing the names that you want to retrieve (see FAQ Advice #12 for details).

Res.:

. mat list A

A[4,1]
              c1
 AMC_Concord   1
   AMC_Pacer   2
  AMC_Spirit   3
Buick_Cent~y   4

.

[Siying Li]

I do not think that's it. This works:

clear
mat A= 1\2\3\4
mat l A
frame create new
frame new{
sysuse auto, clear
gen make2= strtoname(make)
levelsof make2 if _n<5, local(names) clean
}
mat rownames A= `names'
frame drop new
mat list A

As I stated, you probably need to pass the strings through -strtoname()- before running -levelsof-. If this does not help, post a data example using the dataex command of the variable containing the names that you want to retrieve (see FAQ Advice #12 for details).

Res.:

. mat list A

A[4,1]
c1
AMC_Concord 1
AMC_Pacer 2
AMC_Spirit 3
Buick_Cent~y 4

.

Andrew,

Sure! I will try it. Great thanks!