MySQL Couting Distinct values when grouping by dates - mysql

Question

I do have two tables, one is a calendar table with a DATE column, and the other contains ID's and three DATES for each ID.
Calendar Table:
dt
2016-01-01
2016-01-02
2016-01-03
2016-01-04
...
Data Table:
ID d_created d_forwarded d_solved
1 2016-01-01 2016-01-02 2016-01-03
2 2016-01-01 2016-01-02 2016-01-03
3 2016-01-02 2016-01-02 2016-01-04
4 2016-01-03 2016-01-04 2016-01-05
...
The Data Table does in reality contain a multitude of other fields, but I think that is irrelevant for my question. I have a query which selects the DATE field from the calendar table for a given range, let's say a month, and then I do a LEFT JOIN with the Data Table using all three DATE fields combined with OR, because I need to count multiple things from the Data Table, some depending on the d_created, some depending on the d_forwarded and others on d_solved:
SELECT
tc.dt AS dt,
COUNT(DISTINCT(CASE
WHEN td.id != 0
AND DATE(td.d_solved) >= '2016-01-01'
AND DATE(td.d_solved) <= '2016-01-31'
THEN td.id
ELSE NULL
END)) AS result1
.... more stmts ...
FROM calendar_table tc
LEFT JOIN data_table td ON tc.dt = DATE(td.d_created) OR tc.dt = DATE(td.d_solved) OR tc.dt = DATE(td.d_forwarded)
Now here's my problem: The query delivers the correct output, when I do not group my results by tc.dt, but as soon as I group it by tc.dt, the results are incorrect. I am by no means an SQL expert, but as far as I understand it, td.id will occur more than once due to the JOIN, and as long as I have a single result row, the DISTINCT prevents an ID from being counted twice.
I need to be able to count all ID's which have been created, solved or forwarded within my date range, and I also need the calendar table join because I would like to display each day in the range, even though there might be no matching dates in my data table for a particular day, if that makes sense.
Is there any way I can make sure that no ID is counted more than one time when grouping by days ?
I hope I could make clear what the exact problem is, if not, please let me know and I try to elaborate in more detail.
UPDATE
I tried using SarathChandra example which looks quite promising and it does indeed deliver results, however as soon as I add more criteria to my CASE WHEN statement, it does not work the way it should. I forked and modified SarathChandra's ideaone fiddle HERE
So it should return 1 for the 2016-01-02 date but it shows a 0 ?
UPDATE 2
Unfortunately, none of the provided answers was able to solve the underlying problem. While both suggestions were appreciated a lot, I ended up splitting the query into three queries, each time joining the calendar table with the same range of dates, and then combining the arrays in PHP to a single result set.

Answer 1

I have made some assumptions regarding the data in arriving at the following solution:
d_created shall precede d_forwarded, which in turn shall precede d_solved.
In order to remove duplicate counts, that is, count each record only once, I am joining on the basis of the least of the three dates.
The below query seems to be working fine for me.
SELECT
tc.dt AS dt,
COUNT(
CASE WHEN DATE(td.d_created) BETWEEN '2016-01-01' AND '2016-01-31' THEN td.id
ELSE NULL END) AS `Count`
FROM calendar_table tc
LEFT JOIN data_table td ON
(tc.dt = LEAST(DATE(td.d_created), DATE(td.d_solved), DATE(td.d_forwarded)))
GROUP BY tc.dt;
UPDATE: Working example code here.

Answer 2

The problem here is that you join on multiple columns, so when you group on date you'll for example get the ID 1 for the '2016-01-01' (created), '2016-01-02' (solved) and '2016-01-03' (forwarded).
You could try to join to the same table 3 times and count the results in 3 columns. The sum of each column should then match the number of records.
SQL Fiddle Example
Query:
SELECT tc.dt AS dt,
COUNT(DISTINCT(CASE WHEN td_solved.id != 0
AND DATE(td_solved.d_solved) >= '2016-01-01'
AND DATE(td_solved.d_solved) <= '2016-01-31' THEN td_solved.id ELSE NULL END)) AS solved,
COUNT(DISTINCT(CASE WHEN td_created.id != 0
AND DATE(td_created.d_created) >= '2016-01-01'
AND DATE(td_created.d_created) <= '2016-01-31' THEN td_created.id ELSE NULL END)) AS created,
COUNT(DISTINCT(CASE WHEN td_forwarded.id != 0
AND DATE(td_forwarded.d_forwarded) >= '2016-01-01'
AND DATE(td_forwarded.d_forwarded) <= '2016-01-31' THEN td_forwarded.id ELSE NULL END)) AS forwarded
FROM calendar_table tc
LEFT JOIN data_table td_created ON tc.dt = DATE(td_created.d_created)
LEFT JOIN data_table td_solved ON tc.dt = DATE(td_solved.d_solved)
LEFT JOIN data_table td_forwarded ON tc.dt = DATE(td_forwarded.d_forwarded)
GROUP BY 1 WITH ROLLUP