Count the cycle, and count that already has been counted - sql

Question

I have my query:
SELECT UserGroupCode,COUNT(UserGroupCode) AS [CountofCycle]
FROM Users.GroupCycles
GROUP BY UserGroupCode;
Which shows me:
UserGroupCode CountofCycles
1 1
4 1
5 1
6 2 (gone into 2nd cycle)
7 1
8 1
9 1
10 1
11 1
12 1
13 1
14 1
15 1
16 1
17 1
18 1
19 1
When i try to count Total UserGroups where countofcycle=1
SELECT Count(t.CountOfCycle) AS 'totalgroups'
FROM
(SELECT CreateDate, COUNT(userGroupCode) AS [CountofCycle]
FROM Users.GroupCycles
GROUP BY CreateDate,UserGroupCode)t
WHERE CountofCycle=1
I get result = 18 which should be 16, if i delete CreateDate from both SELECT And GROUP BY statement i can get correct number of CountofCycles,
and when i change condition to CountofCycle=2 or >1 it shows me 0
What is the problem with showing UserGroups with cycle > 1 ???!??
Here is my query to filter out onCreateDate, in 2nd table that i UNION with 1st one, i cant't use CreateDate, as it disturbs my query results
SELECT Count(t.CountOfCycle) AS 'total groups'
FROM
(SELECT COUNT(userGroupCode) AS [CountofCycle], CreateDate
FROM users.GroupCycles GROUP BY userGroupCode,CreateDate)t
WHERE t.CountOfCycle=1 AND t.CreateDate Between '03/16/2017' AND '04/25/2017'
UNION ALL
SELECT Count(t.CountOfCycle) AS 'group on date2'
FROM
(SELECT COUNT(userGroupCode) AS [CountofCycle] FROM users.GroupCycles GROUP BY userGroupCode)t
WHERE t.CountOfCycle=2

Answer 1

Firstly to address why you are not getting the results you are expecting, and the simple reason is that you are comparing two different queries and expecting the results to be the same.
Consider this very simple example data
UserGroupCode | CreateDate
----------------+----------------
A | 2017-05-10
B | 2017-05-10
B | 2017-05-11
C | 2017-05-10
You have two records where the UserGroupCode is B, so if you run:
DECLARE #T TABLE (UserGroupCode CHAR(1), CreateDate DATE)
INSERT #T (userGroupCode, CreateDate)
VALUES ('A', '2017-05-10'), ('B', '2017-05-10'), ('B', '2017-05-11');
SELECT UserGroupCode, COUNT(*) AS [Count]
FROM #T
GROUP BY UserGroupCode
HAVING COUNT(*) = 2;
This returns:
UserGroupCode Count
-------------------------
B 2
However, if you were to add CreateDate to the grouping, "B" would be split into two groups, each with a count of 1:
DECLARE #T TABLE (UserGroupCode CHAR(1), CreateDate DATE)
INSERT #T (userGroupCode, CreateDate)
VALUES ('A', '2017-05-10'), ('B', '2017-05-10'), ('B', '2017-05-11');
SELECT UserGroupCode, CreateDate, COUNT(*) AS [Count]
FROM #T
WHERE UserGroupCode = 'B'
GROUP BY UserGroupCode, CreateDate;
This returns:
UserGroupCode CreateDate Count
---------------------------------------
B 2017-05-10 1
B 2017-05-11 1
Now, based you your queries you have posted, it looks like you want to know
The number of groups that only have one record in the date range 16th March 2017 to 25th April 2017.
The number of groups that have two records in total.
For this, consider a slightly larger data set:
UserGroupCode | CreateDate
----------------+----------------
A | 2017-04-10
B | 2017-04-10
B | 2017-05-11
C | 2017-01-01
C | 2017-01-02
D | 2017-04-01
D | 2017-04-02
E | 2017-01-02
So here.
Group A has one record in total, and it falls within the date range
Group B has two records in total, on in the date range, one not
Group C has two records, neither in the date range
Group D has two records, both in the date range.
Group E has one record, not in the date range
So for your first requirement:
The number of groups that only have one record in the date range 16th March 2017 to 25th April 2017.
We would expect 2 groups, A and B, because C and E have no records in the date range, and D has two.
and the second we would expect three groups, B, C and D, since A and E only have one record each.
You can do this with a single query by using a conditional aggregate.
DECLARE #T TABLE (UserGroupCode CHAR(1), CreateDate DATE)
INSERT #T (userGroupCode, CreateDate)
VALUES ('A', '2017-04-10'),
('B', '2017-04-10'), ('B', '2017-05-11'),
('C', '2017-01-01'), ('C', '2017-01-02'),
('D', '2017-04-01'), ('D', '2017-04-02'),
('E', '2017-01-02');
SELECT TotalGroups = COUNT(CASE WHEN RecordsInPeriod = 1 THEN 1 END),
GroupOnDate2 = COUNT(CASE WHEN TotalRecords = 2 THEN 1 END)
FROM ( SELECT UserGroupCode,
TotalRecords = COUNT(*),
RecordsInPeriod = COUNT(CASE WHEN CreateDate >= '20170316'
AND CreateDate <= '20170425' THEN 1 END)
FROM #T
GROUP BY UserGroupCode
) AS t;
Which gives:
TotalGroups GroupOnDate2
------------------------------
2 3

Answer 2

I'd expect to see a HAVING clause rather than a WHERE:
SELECT UserGroupCode, COUNT(UserGroupCode) [CountofCycle]
FROM [Users].[GroupCycles]
GROUP BY UserGroupCode
HAVING COUNT(UserGroupCode) > 1;

Answer 3

You could use HAVING, should work (and be more efficient)
select count(*)
from
(
SELECT CreateDate, COUNT(userGroupCode) AS [CountofCycle]
FROM Users.GroupCycles
GROUP BY CreateDate,UserGroupCode
having count(userGroupCode) > 1 -- here is HAVING clause
) x1