Running standard deviation to date

[Marcus Colby]

Hi all,

First time post so bare with me if I've left out any critical details.
I'm using STATA version 12.1 on Mac.

I'm trying to calculate by ID, Season, and Phase the running standard deviation up to any given date.

Example dummy data below. There is not a training session on every day hence the missing data and I only want to include dates where there was a session in the standard deviation calculation. If there is a change in ID, Season, or Phase I would like the standard deviation calculation to reset and start again.

Example 1, for ID 1 on the 8/1/16 I would like the SD from 1/1/16 to the 7/1/16. Example 2, for ID 1 on the 17/1/16 I would like the SD from 1/1/16 to the 16/1/16.

It's a time series dataset and therefore I set it accordingly:

tsset ID Date
tsfill

Date	ID	Season	Phase	Distance
1/1/16	1	2016	1	8000
2/1/16	1	2016	1	.
3/1/16	1	2016	1	4000
4/1/16	1	2016	1	.
5/1/16	1	2016	1	3000
6/1/16	1	2016	1	.
7/1/16	1	2016	1	.
8/1/16	1	2016	1	2000
9/1/16	1	2016	1	.
10/1/16	1	2016	1	.
11/1/16	1	2016	1	7000
12/1/16	1	2016	1	.
13/1/16	1	2016	1	.
14/1/16	1	2016	1	9000
15/1/16	1	2016	1	.
16/1/16	1	2016	1	.
17/1/16	1	2016	1	8000
18/1/16	1	2016	1	.
19/1/16	1	2016	1	4000
20/1/16	1	2016	2	.
21/1/16	1	2016	2	3000
22/1/16	1	2016	2	.
23/1/16	1	2016	2	.
24/1/16	1	2016	2	2000
25/1/16	1	2016	2	.
26/1/16	1	2016	2	.
27/1/16	1	2016	2	7000
28/1/16	1	2016	2	.
29/1/16	1	2016	2	.
30/1/16	1	2016	2	7000
31/1/16	1	2016	2	.
1/2/16	1	2016	2	8000
2/2/16	1	2016	2	.
3/2/16	1	2016	2	.
4/2/16	1	2016	2	9000
5/2/16	1	2016	2	.
1/1/16	2	2016	1	8000
2/1/16	2	2016	1	.
3/1/16	2	2016	1	4000
4/1/16	2	2016	1	.
5/1/16	2	2016	1	3000
6/1/16	2	2016	1	.
7/1/16	2	2016	1	.
8/1/16	2	2016	1	2000
9/1/16	2	2016	1	.
10/1/16	2	2016	1	.
11/1/16	2	2016	1	7000

Any help on this would be greatly appreciated!

[Nick Cox]

Please use dataex (SSC) to show examples, as outlined in the FAQ Advice #12. Dates are a particular pain, as is shown by the editing needed to produce an example that will run.

Otherwise this appears to yield to rangestat (SSC): search the forum for mentions. -1000 is an arbitrarily large first argument.

clear 
input str8 sDate    ID    Season    Phase    Distance
"1/1/16"    1    2016    1    8000
"2/1/16"    1    2016    1    .
"3/1/16"    1    2016    1    4000
"4/1/16"    1    2016    1    .
"5/1/16"    1    2016    1    3000
"6/1/16"    1    2016    1    .
"7/1/16"    1    2016    1    .
"8/1/16"    1    2016    1    2000
"9/1/16"    1    2016    1    .
"10/1/16"    1    2016    1    .
"11/1/16"    1    2016    1    7000
"12/1/16"    1    2016    1    .
"13/1/16"    1    2016    1    .
"14/1/16"    1    2016    1    9000
"15/1/16"    1    2016    1    .
"16/1/16"    1    2016    1    .
"17/1/16"    1    2016    1    8000
"18/1/16"    1    2016    1    .
"19/1/16"    1    2016    1    4000
"20/1/16"    1    2016    2    .
"21/1/16"    1    2016    2    3000
"22/1/16"    1    2016    2    .
"23/1/16"    1    2016    2    .
"24/1/16"    1    2016    2    2000
"25/1/16"    1    2016    2    .
"26/1/16"    1    2016    2    .
"27/1/16"    1    2016    2    7000
"28/1/16"    1    2016    2    .
"29/1/16"    1    2016    2    .
"30/1/16"    1    2016    2    7000
"31/1/16"    1    2016    2    .
"1/2/16"    1    2016    2    8000
"2/2/16"    1    2016    2    .
"3/2/16"    1    2016    2    .
"4/2/16"    1    2016    2    9000
"5/2/16"    1    2016    2    .
"1/1/16"    2    2016    1    8000
"2/1/16"    2    2016    1    .
"3/1/16"    2    2016    1    4000
"4/1/16"    2    2016    1    .
"5/1/16"    2    2016    1    3000
"6/1/16"    2    2016    1    .
"7/1/16"    2    2016    1    .
"8/1/16"    2    2016    1    2000
"9/1/16"    2    2016    1    .
"10/1/16"    2    2016    1    .
"11/1/16"    2    2016    1    7000
end 
gen Date = daily(sDate, "DMY", 2050) 

rangestat (sd) Distance, interval(Date -1000 -1) by(ID Season Phase)

[Marcus Colby]

Nick, you're a great man. Thank you very much!