如何根据另一列的值过滤一列？

Question

I apologize in advanced for the long post. It's a couple of case statements that make it really long.

I have two questions. The first is -- how do I get all the values from a table, but also filter it against itself in a self join? Specifically, I am counting the number of calls salespeople made per day and my output keeps cutting off 12am and 1am.

The second question is -- how do I filter one column based on the values of another column?

Some of my salespeople are in New York, so I want to say when created_by LIKE 'Johnny Smith' THEN subtract 3 from TIME. No clue how to do that!

Here's my example--

This is what my query looks like...

SELECT  CASE
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 1 THEN '1am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 2 THEN '2am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 3 THEN '3am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 4 THEN '4am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 5 THEN '5am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 6 THEN '6am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 7 THEN '7am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 8 THEN '8am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 9 THEN '9am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 10 THEN '10am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 11 THEN '11am'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 12 THEN '12pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 13 THEN '1pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 14 THEN '2pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 15 THEN '3pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 16 THEN '4pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 17 THEN '5pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 18 THEN '6pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 19 THEN '7pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 20 THEN '8pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 21 THEN '9pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 22 THEN '10pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 23 THEN '11pm'
           WHEN (EXTRACT(HOUR
                         FROM task1.time)) = 0 THEN '12am'
       END AS hour,
       COUNT(task2.Activity_ID__Case_Safe_) AS calls
FROM sdr_data.by_task AS task1
LEFT JOIN sdr_data.by_task AS task2
  ON task1.Activity_ID__Case_Safe_ = task2.Activity_ID__Case_Safe_
  AND task2.task_type LIKE 'Call'
  AND task2.status LIKE 'Completed'
  AND task2.created_date > DATE_SUB(CURRENT_DATE(), INTERVAL 1
                                    WEEK)
LEFT JOIN
  (SELECT account_id__CaseSafe_
   FROM sdr_data.by_task
   WHERE account_id__CaseSafe_ IN
       (SELECT account_id__CaseSafe_
        FROM {{@dummy_accounts}})) AS subquery
  ON task2.account_id__CaseSafe_ = subquery.account_id__CaseSafe_
RIGHT JOIN
  (SELECT created_by,
          Activity_ID__Case_Safe_
   FROM sdr_data.by_task
   WHERE created_by IN
       (SELECT full_name
        FROM {{@active_salespeople}}) ) AS subquery2
  ON task1.Activity_ID__Case_Safe_ = subquery2.Activity_ID__Case_Safe_
WHERE task1.created_by IS NOT NULL
  AND subquery.account_id__CaseSafe_ IS NULL
GROUP BY 1
ORDER BY CASE
             WHEN HOUR LIKE '12am' THEN 1
             WHEN HOUR LIKE '1am' THEN 2
             WHEN HOUR LIKE '2am' THEN 3
             WHEN HOUR LIKE '3am' THEN 4
             WHEN HOUR LIKE '4am' THEN 5
             WHEN HOUR LIKE '5am' THEN 6
             WHEN HOUR LIKE '6am' THEN 7
             WHEN HOUR LIKE '7am' THEN 8
             WHEN HOUR LIKE '8am' THEN 9
             WHEN HOUR LIKE '9am' THEN 10
             WHEN HOUR LIKE '10am' THEN 11
             WHEN HOUR LIKE '11am' THEN 12
             WHEN HOUR LIKE '12pm' THEN 13
             WHEN HOUR LIKE '1pm' THEN 14
             WHEN HOUR LIKE '2pm' THEN 15
             WHEN HOUR LIKE '3pm' THEN 16
             WHEN HOUR LIKE '4pm' THEN 17
             WHEN HOUR LIKE '5pm' THEN 18
             WHEN HOUR LIKE '6pm' THEN 19
             WHEN HOUR LIKE '7pm' THEN 20
             WHEN HOUR LIKE '8pm' THEN 21
             WHEN HOUR LIKE '9pm' THEN 22
             WHEN HOUR LIKE '10pm' THEN 23
             WHEN HOUR LIKE '11pm' THEN 24
         END

For context, the first join is to filter out the tasks so I just get completed calls. The second join is to get rid of bad/fake accounts and the third join is to make sure it only pulls names from the "active-salespeople" list.

Here's an example of what my source data looks like --

Created By  Task Type   Time    Status  Activity ID (Case Safe)
Rudolph Ronald  Call    2015-12-21 12:56:00 Completed   61AA
Joey Jalla  Call    2015-12-21 12:56:00 Completed   61BB
Mike McCrystal  Call    2015-12-21 12:56:00 Completed   61CC
Sally Susie Email   2015-12-21 12:56:00 Open    61DD
Sally Susie Email   2015-12-21 12:56:00 Open    61EE
Sally Susie Meeting 2015-12-21 12:56:00 Completed   61FF

My output looks like this --

hour    calls
2am 5
3am 14
4am 8
5am 20
6am 52
7am 55
8am 145
9am 90
10am    106
11am    93
12pm    96
1pm 53
2pm 54
3pm 43
4pm 20
5pm 8
6pm 4
7pm 17
8pm 9
9pm 12
11pm    6

But I want it to look like --

hour    calls
12am    2
1am 4
2am 5
3am 14
4am 8
5am 20
6am 52
7am 55
8am 145
9am 90
10am    106
11am    93
12pm    96
1pm 53
2pm 54
3pm 43
4pm 20
5pm 8
6pm 4
7pm 17
8pm 9
9pm 12
11pm    6

Answer 1

SELECT DATE_FORMAT(task1.time, "%l%p") hour,
       COUNT(task2.Activity_ID__Case_Safe_) AS calls

 YOUR REST OF THE QUERY 

GROUP BY 1
ORDER BY hour

1) You can replace the case statement with

DATE_FORMAT(task1.time, "%l%p") ---This gives hour in AM/PM

And also order by alias rather than writing the logic in order by

2) In your query there is a condition in the outer query , (subquery.account_id__CaseSafe_ IS NULL)

But case safe is not null for the time 12 AM in the sample data. So it is getting filtered out.

3) If you want to change time across timezone then use

if name = "Johnny Smith", 
     DATE_FORMAT(date_add(task1.time,  INTERVAL 3 HOUR) + 3, "%l%p") , 
     DATE_FORMAT(task1.time + 3, "%l%p"))