Average over intersect - sql

Question

I've got this data set where I'm creating a report per shift displaying a certain activity per hour. Eg.: A shift runs from 7-15, so the report will be:
7:00 - 15 moves
8:00 - 18 moves
This report contains filters for weeks and days, as the goal is for users to see their shift performance.
However, I'd like to compare the activities in a specific hour with the same hour, same day in the last 10 weeks. So for example:
I've got 15 moves in hour 7:00 of a Tuesday and i'd like to compare that with the average move count on 7:00 of a Tuesday in the last 10 weeks.
How would I get this in SQL! :-)

Answer 1

Here is something that might get you moving in the right direction. Given that you have not stated your data structure, I have taken the liberty to assume what that may be.
You can run the following in SSMS to review/modify the results yourself.
First, I created a table variable to simulate the assumed data.
-- Create a dummy "shift" table --
DECLARE #shifts TABLE ( shift_id INT IDENTITY ( 1, 1 ) PRIMARY KEY, shift_date DATETIME, shift_moves INT );
Then I inserted some dummy data. For the sake of time and simplicity, I stuck with Tuesdays over the last ten weeks.
-- Insert ten weeks of Tuesdays --
INSERT INTO #shifts ( shift_date, shift_moves ) VALUES
( '11/06/2018 07:10:00', 5 )
, ( '11/13/2018 07:08:00', 12 )
, ( '11/20/2018 07:00:00', 14 )
, ( '11/27/2018 07:20:00', 15 )
, ( '12/04/2018 07:35:00', 12 )
, ( '12/11/2018 07:18:00', 11 )
, ( '12/18/2018 07:16:00', 10 )
, ( '12/25/2018 07:00:00', 12 )
, ( '01/01/2019 07:00:00', 13 )
, ( '01/08/2019 07:22:00', 15 );
The table variable #shifts now contains the following data:
+----------+-------------------------+-------------+
| shift_id | shift_date | shift_moves |
+----------+-------------------------+-------------+
| 1 | 2018-11-06 07:10:00.000 | 5 |
| 2 | 2018-11-13 07:08:00.000 | 12 |
| 3 | 2018-11-20 07:00:00.000 | 14 |
| 4 | 2018-11-27 07:20:00.000 | 15 |
| 5 | 2018-12-04 07:35:00.000 | 12 |
| 6 | 2018-12-11 07:18:00.000 | 11 |
| 7 | 2018-12-18 07:16:00.000 | 10 |
| 8 | 2018-12-25 07:00:00.000 | 12 |
| 9 | 2019-01-01 07:00:00.000 | 13 |
| 10 | 2019-01-08 07:22:00.000 | 15 |
+----------+-------------------------+-------------+
I created a few parameters that you might pass to a stored procedure.
-- What date are we looking at? --
DECLARE #date DATETIME = '01/08/2019';
-- How many weeks back to compare? --
DECLARE #weeks_back INT = -10;
*Remember: To look backward, #weeks_back MUST be a negative number. In production, you would have a check/handle for this.
Next, I created two local variables to help keep the date/time simple for use when querying.
-- Create variables for the start and end times for simplicity --
DECLARE
#sDT DATETIME = CAST( CONVERT( VARCHAR(10), #date, 101 ) + ' 00:00:00' AS DATETIME ),
#eDT DATETIME = CAST( CONVERT( VARCHAR(10), #date, 101 ) + ' 23:59:59' AS DATETIME );
Then, it was a matter of querying the data for the desired results using CROSS APPLY.
Using CROSS APPLY allows me to query a second subset of shift data that is tied to the primary record queried ( in this case a shift_date of 01/08/2019 ) for the desired time frame ( #weeks_back ).
-- Get resultset --
SELECT
DATEPART( hh, shift_date ) AS [shift_hour]
, DATENAME( dw, s.shift_date ) AS [shift_day]
, CONVERT( VARCHAR(10), s.shift_date, 101 ) AS [shift_date]
, s.shift_moves
, shift_avg.shift_average
FROM #shifts AS s
CROSS APPLY (
SELECT
AVG( a.shift_moves ) AS [shift_average]
FROM #shifts a
WHERE
-- restrict to the current hour.
DATEPART( HH, a.shift_date ) = DATEPART( HH, s.shift_date )
-- restrict to the current day of the week.
AND DATEPART( DW, a.shift_date ) = DATEPART( DW, s.shift_date )
-- compare against the desired time period / weeks back.
AND a.shift_date BETWEEN DATEADD( WW, #weeks_back, CAST( CONVERT( VARCHAR(10), s.shift_date, 101 ) AS DATETIME ) ) AND a.shift_date
) AS shift_avg
WHERE
s.shift_date BETWEEN #sDT AND #eDT
ORDER BY
s.shift_date;
Which returns the following resultset:
+------------+-----------+------------+-------------+---------------+
| shift_hour | shift_day | shift_date | shift_moves | shift_average |
+------------+-----------+------------+-------------+---------------+
| 7 | Tuesday | 01/08/2019 | 15 | 11 |
+------------+-----------+------------+-------------+---------------+
I hope this helps get you moving in the desired direction, #Aron.