Counting Occurances of an Event in a Day - tsql

Question

Looking for a real quick and dirty dump of data from a table that more or less has this layout...
ID | EventType | EventDate
------+-------------+------------
1 | Inbound | 2018-07-18 00:00:00
2 | Outbound | 2018-07-18 12:00:00
3 | Inbound | 2018-07-19 00:12:00
4 | Failure | 2018-07-19 00:12:00
5 | Inbound | 2018-07-19 00:12:00
6 | Outbound | 2018-07-19 00:12:00
| |
And what I want out of it is a query that spits out the count of each occurrence for a day. So '2018-07-19' would spit me out
Failures | Inbounds | Outbounds
-----------+----------+------------
1 | 2 | 1
Here's my real crap attempt at it, but I assume there's an easier way to get away with it. Ideally I'd be able to drop this in a view and filter by date on my own but if I need to pass a target date on stored proc then that's fine I suppose.
There's only 3 defined event types on my database so my static solution of counting each one is fine. Having something that dynamically adapts to however many distinct event types would be better but not necessary
DECLARE #TestDate datetime2 = '2018-07-19 08:41:55'
SELECT
'Failures' = SUM(Failures),
'Inbounds' = SUM(Inbounds),
'Outbounds'= SUM(Outbounds)
FROM (
SELECT 'Failures' = COUNT(ID), 'Inbounds' = 0, 'Outbounds' = 0 FROM tblTests WHERE EventType = 'Failed' AND EventDate BETWEEN CAST(#TestDate AS DATE) AND DATEADD(DAY, 1, CAST(#TestDate AS DATE)) UNION
SELECT 'Failures' = 0, 'Inbounds' = COUNT(ID), 'Outbounds' = 0 FROM tblTests WHERE EventType = 'Inbound' AND EventDate BETWEEN CAST(#TestDate AS DATE) AND DATEADD(DAY, 1, CAST(#TestDate AS DATE)) UNION
SELECT 'Failures' = 0, 'Inbounds' = 0, 'Outbounds' = COUNT(ID) FROM tblTests WHERE EventType = 'Outbound' AND EventDate BETWEEN CAST(#TestDate AS DATE) AND DATEADD(DAY, 1, CAST(#TestDate AS DATE))
) FIO

Answer 1

The easy way to do such things is called conditional aggregating:
SELECT Cast(EventDate As Date) As SummaryDate,
SUM(CASE WHEN EventType = 'Failed' THEN 1 ELSE 0 END) As Failures,
SUM(CASE WHEN EventType = 'Inbound' THEN 1 ELSE 0 END) As Inbounds,
SUM(CASE WHEN EventType = 'Outbound' THEN 1 ELSE 0 END) As Outbounds
FROM tblTests
GROUP BY Cast(EventDate As Date)

Answer 2

This would be a great opportunity in my opinion to use a pivot with a count aggregate.
The example below walks through creating a test table loading it with the data from the OP original question and pivoting the results on date
Create the test table
create table testtable (id int, value varchar(20), dt datetime)
Load the temp data into the new table
insert into testtable
values( 1, 'Inbound', '2018-07-18 00:00:00'),
(2, 'Outbound', '2018-07-18 12:00:00'),
(3, 'Inbound' , '2018-07-19 00:12:00'),
(4, 'Failure' , '2018-07-19 00:12:00'),
(5, 'Inbound' , '2018-07-19 00:12:00'),
(6, 'Outbound' , '2018-07-19 00:12:00')
Pivot the data to the correct result
select * from (
select value, cast(dt as date) d
from testtable )a
pivot(
count(value) for value in ([Inbound],[Outbound],[Failure]))piv
This returns results this as a result
This can easily be expanded by adding additional values into the pivot.

Answer 3

Here's what you need
SELECT EventType, COUNT(1) AS Cnt, CONVERT(DATE, EventDate) AS EventDate
FROM dbo.tblTests
GROUP BY EventType, CONVERT(DATE, EventDate)
You can put a WHERE to limit the results to a specific date.
If you plan to run this often and run it on a large table I'll suggest that you create a computed persisted column to have the date only,
You can do this like that:
ALTER TABLE tblTests ADD DateOnly AS CONVERT(DATE, EventDate) PERSISTED NOT NULL
Then use DateOnly in place of CONVERT(DATE, EventDate) in the above query.
And for best performance you can create an index on EventType and the new DateOnly columns
like that:
CREATE NONCLUSTERED INDEX [NCI_tblTest__EventType__DateOnly] ON [dbo].[tblTests]
(
[EventType] ASC,
[DateOnly ] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, SORT_IN_TEMPDB = OFF, DROP_EXISTING = OFF, ONLINE = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON)
and remember to rebuild indexes regularly.
EDIT:
This will give you different event types on a separate rows.
EDIT2: use PIVOT to produce the results in the exact same way as you want it. Leaving that for you to exercise :) https://learn.microsoft.com/en-us/sql/t-sql/queries/from-using-pivot-and-unpivot?view=sql-server-2017