Counting Number of Times a Code Takes A Different Value

[Brooke Claypole]

Hello,

I have been working with stata for my dissertation and I consider myself a beginner. I am using Stata MP/17.0. I am attempting to create a new variable (tot_chg_code) that counts the number of times a code takes a different value for each ID. Variables of interest are ID, code, date. date is format %td. My dataset has 30,634 unique IDs and 222,381 observations. There are more than 99 unique values for code. Dates range over 30 years. I have searched the list and several help files, but I am not sure how to achieve what I am looking for. I know I need to sort the data by ID and date before I attempt to calculate tot_chg_code. I have created a simple example of my data. Dataex below.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(ID code date)
1 19 11868
2 29 14020
2 39 15063
2 49 16245
2 49 16265
2 59 16289
2 59 16845
2 69 16888
2 59 17691
2 49 17701
2 59 17735
end
format %td date

Once the value tot_chg_code is calculated, I will collapse the data by ID. I would like to end up with:

ID	tot_chg_code
1	0
2	7

Anyone have suggestions on code I might try? Thank you.

[Clyde Schechter]

* Example generated by -dataex-. For more info, type help dataex
clear
input float(ID code date)
1 19 11868
2 29 14020
2 39 15063
2 49 16245
2 49 16265
2 59 16289
2 59 16845
2 69 16888
2 59 17691
2 49 17701
2 59 17735
end
format %td date

by ID (date), sort: gen wanted = sum(code != code[_n-1])
by ID (date): replace wanted = wanted[_N] - 1

To get to your final result, use -collapse- or -by ID: keep if _n == 1-.

[Brooke Claypole]

Thank you, Clyde Schechter. This works great!