Question for data modelling experts [Sample data p... - Qlik Community

mwallman · ‎2018-05-11

Hello data experts,

I have the following example of data where I am unable find out exactly how to model it properly.

The goal

The goal is to analyse how long patient's have walked per day as well as how long they have slept.

The data (sample):

Activity:

Date	Calories Burned	Distance
01-04-2018	2,976	6.82
02-04-2018	3,060	7.89
03-04-2018	2,570	7.65
04-04-2018	2,591	7.8
05-04-2018	2,615	8.03
06-04-2018	2,700	32.41
07-04-2018	4,385	27.64
08-04-2018	3,005	10.73
09-04-2018	2,538	23.55
10-04-2018	2,682	34.78
11-04-2018	2,693	54.89
12-04-2018	2,636	24.04
13-04-2018	3,327	11
14-04-2018	3,402	11.06
15-04-2018	3,304	10.36
16-04-2018	2,552	3.66
17-04-2018	2,963	5.89
18-04-2018	2,517	13.61
19-04-2018	2,566	33.9
20-04-2018	2,846	17.23
21-04-2018	3,196	19.07
22-04-2018	3,105	18.81
23-04-2018	2,690	14.78

Sleep:

Start Time	End Time	Minutes Asleep
29-04-2018 10:11 pm	30-04-2018 6:22 am	415
28-04-2018 11:19 pm	29-04-2018 7:28 am	429
28-04-2018 12:38 am	28-04-2018 4:51 am	237
25-04-2018 11:09 pm	26-04-2018 8:13 am	515
24-04-2018 11:55 pm	25-04-2018 7:21 am	382
23-04-2018 11:33 pm	24-04-2018 6:40 am	440
22-04-2018 11:48 pm	23-04-2018 6:21 am	389
21-04-2018 12:50 pm	22-04-2018 8:57 am	474
19-04-2018 11:09 pm	20-04-2018 5:57 am	379
18-04-2018 11:25 pm	19-04-2018 6:41 am	383
17-04-2018 11:40 pm	18-04-2018 6:47 am	414
16-04-2018 10:53 pm	17-04-2018 5:07 am	333
16-04-2018 12:46 am	16-04-2018 6:27 am	306
14-04-2018 11:45 pm	15-04-2018 7:29 am	404
14-04-2018 12:02 am	14-04-2018 7:46 am	422
12-04-2018 11:35 pm	13-04-2018 7:24 am	425
11-04-2018 11:20 pm	12-04-2018 6:39 am	370
10-04-2018 11:01 pm	11-04-2018 7:11 am	425
09-04-2018 10:44 pm	10-04-2018 6:51 am	431
08-04-2018 11:14 pm	09-04-2018 6:23 am	387
08-04-2018 12:42 am	08-04-2018 8:31 am	423
06-04-2018 11:56 pm	07-04-2018 7:29 am	425
05-04-2018 11:30 pm	06-04-2018 7:30 am	443
05-04-2018 12:04 am	05-04-2018 7:00 am	356
03-04-2018 11:06 pm	04-04-2018 6:22 am	392
03-04-2018 12:43 am	03-04-2018 6:36 am	300
01-04-2018 09:37 pm	02-04-2018 8:16 am	470
31-03-2018 10:11 pm	01-04-2018 10:15 am	646

Problem:

The problem I have is the only thing that links the two data sets above is dates.

It is easy to identify steps taken per day in the Activity table. And you can get the date from the End Date in the Sleep table.

But the sleep data is problematic because I am not sure how to figure out the following logic:

If the Start Time is Yesterday and End Time is today then sum Time Asleep and record it for the value from the End Date
If the Start Time and End Time are in the same date then record it for the value from the End Date
If the are multiple entries (i.e. for naps) then also enter it for the value from the End Date.

Anyone know how to transform the Sleep data that works for the above logic?

petter · ‎2018-05-13

I think the DayStart() function can be a solution for you. It can take a dayshift parameter into account and if you specify 0.5 which is half a day or 12 hours as a dayshift it might be exactly what you need. From your example data I see that you never sleep past noon and you never start sleeping for the new day until several hours after noon...

[ACTIVITY]:
LOAD
 Date(Date#([Date], 'DD-MM-YYYY') ) AS [Date],
 [Calories Burned],
 [Distance]
 FROM [lib://ActivityAndSleep]
(html, utf8 , embedded labels, table is @1);


[SLEEP]:
LOAD
    *,
    Floor(DayStart([Start Time],0,0.5)) AS Date;
LOAD
 Timestamp(Timestamp#([Start Time], 'DD-MM-YYYY hh:mm TT') ) AS [Start Time],
 Timestamp(Timestamp#([End Time], 'DD-MM-YYYY h:mm TT') ) AS [End Time],
 [Minutes Asleep]
 FROM [lib://ActivityAndSleep]
(html, utf8 , embedded labels, table is @2);

Note line number 12 where the DayStart() has been used.

Question for data modelling experts [Sample data provided]: How to analyse Start and End where the end time may be next day or on the same day?