Problem with count variable generation

[Sally Hansen]

Dear Statalist Users,

Although I went through many related discussions and explanations about my question, nothing helped me. I will be thankful for your suggestions in the following problem.
I am working with a Call Details Record (CDR) Dataset. In the data, for each caller id I observe the antenna used to make/receive call. I want to generate the variable for the number of times a certain antenna is used by a caller. What I am looking for is the following:



Thanks in advance for your cooperation. Bests.


S.

[Sally Hansen]

caller_id	time	antenna_id	count_var
1	t_11	50	2
1	t_12	55	1
1	t_13	50	2
2	t_21	51	1
2	t_22	52	1
2	t_23	53	3
2	t_24	53	3
2	t_25	53	3
3	t_31	55	2
3	t_32	55	2
3	t_33	50	1
3	t_34	51	1
3	t_35	52	1
3	t_36	54	3
3	t_37	54	3
3	t_38	54	3

[Chen Samulsion]

you can try the following code:

bysort caller_id antenna_id: egen count=count( antenna_id )

and for preserving original order of data, you can try:

gen id=_n
bysort caller_id antenna_id: egen count=count( antenna_id )
sort id

[Joro Kolev]

What Chen said.

I would just clarify that Chen's second piece of code that preserves the sort is unnecessary. One can just use the syntax below.

According to Nick Cox the following syntax is no longer documented, and it is no longer explained in the manuals that the following syntax is sort preserving:

.  egen count=count( antenna_id ), by(caller_id antenna_id)

[Chen Samulsion]

Joro Kolev Thank you for introducing this wonderful legacy.

[Sally Hansen]

Dear Chen and Joro, thank you so much for your kind help and suggestions. Both codes (1 and 3) are giving the same outcome and they are exactly doing what I need to do. Bests

S.