Combine a bunch of dummy variables into a multinomial/categorical

[Pratap Pundir]

It is trivial to convert a categorical variable into a bunch of dummy variables.

I would like to do the reverse...I have about 200 dummy variables which I would like to convert into a single categorical variable, so that I can easily run tab etc commands on it.

Any suggestions?

Alternatively, can I run some command on a group of dummies...which has the same effect as running tab on the categorical I would have obtained by combining the dummies.?

e.g.
I have dunmies cat1, cat2, cat3

The output I seek is:
cat1 : 45
cat2 : 36
cat3: 99

[Fei Wang]

An example code as below: convert categorical variable rep78 into individual dummies (r1-r5) and then combine the dummies back into a single categorical variable r.

sysuse auto, clear
tab rep78, generate(r)
gen r = .

forvalues i = 1/5 {
    replace r = `i' if r`i' == 1
}

tab rep78 r, m

[Carlo Lazzaro]

Pratap:
do you mean something along the following lines?

. set obs 3
number of observations (_N) was 0, now 3

. g id=_n

. g cat_1=1 in 2/3

. replace cat_1=0 if cat_1==.

. g cat_2=1 in 1/2

. replace cat_2=0 if cat_2==.

. g cat_3=1 in 1/2

. replace cat_3=0 if cat_3==.

. egen wanted=rowtotal(cat_*)

. list

     +-------------------------------------+
     | id   cat_1   cat_2   cat_3   wanted |
     |-------------------------------------|
  1. |  1       0       1       1        2 |
  2. |  2       1       1       1        3 |
  3. |  3       1       0       0        1 |
     +-------------------------------------+

.

[George Ford]

would egen group work? And do you need to know what the categories mean?