Reputation: 346
I need to make a SQL query
table 'records' structure:
contact_id(integer),
client_id(integer),
worker_id(integer),
statement_status(varchar),
contact_ts(timestamp)
It has to show the following:
current date
number of clients which last statement_status was 'interested'
number of clients which last statement_status was 'not_interested' and previus status was 'not_present'
Could somebody help?
sample data:
contact_id client_id contact_ts worker_id statement_status
'1', '181', '2017-09-24 03:38:31.000000', '107', 'voicemail'
'2', '72', '2017-09-23 09:32:38.000000', '10', 'not_interested'
'3', '277', '2017-09-22 07:06:16.000000', '119', 'interested'
'4', '36', '2017-09-21 04:39:57.000000', '118', 'not_present'
'5', '33', '2017-09-20 04:12:12.000000', '161', 'voicemail'
'6', '244', '2017-09-19 02:26:30.000000', '13', 'not_interested'
'7', '346', '2017-09-18 02:30:35.000000', '255', 'interested'
'8', '128', '2017-09-17 06:20:13.000000', '52', 'not_present'
'9', '33', '2017-09-16 08:58:02.000000', '188', 'not_present'
'10', '352', '2017-09-15 08:18:40.000000', '324', 'not_interested'
'11', '334', '2017-09-14 04:27:40.000000', '373', 'interested'
'12', '2', '2017-09-13 08:44:40.000000', '40', 'not_present'
'13', '33', '2017-09-12 03:46:16.000000', '252', 'voicemail'
'14', '366', '2017-09-11 04:31:22.000000', '78', 'not_interested'
'15', '184', '2017-09-10 06:08:01.000000', '289', 'interested'
'16', '184', '2017-09-09 05:45:56.000000', '124', 'not_present'
'17', '102', '2017-09-08 07:09:30.000000', '215', 'voicemail'
'18', '140', '2017-09-07 08:09:18.000000', '196', 'not_interested'
'19', '315', '2017-09-06 05:13:40.000000', '242', 'interested'
'20', '268', '2017-09-05 07:41:40.000000', '351', 'not_present'
'21', '89', '2017-09-04 05:32:05.000000', '232', 'voicemail'
desired output:
Time, interested, not-interested
2017-09-10 06:08:01, 5, 5
I tried something with sub queries, but it obviously doesn't work:
SELECT
GETDATE()
,(select count(*)
from record a
where (select statement_status
from record
where client_id == a.client_id
order by a.contact_ts
limit 1) == "interested"
group by a.contact_id)
,(select count(*)
from record a
where (select (select statement_status
from record
where client_id == a.client_id
order by a.contact_ts
limit 2) order by a.contact_ts desc limit 1) == "interested"
and
(select statement_status
from record
where client_id == a.client_id
order by a.contact_ts
limit 1) == "interested"
group by a.contact_id)
from record b;
How should I use the inner selects?
I must write a poem, because most of my post is a code.
So maybe something from "Dead man"?
“Don't let the sun burn a hole in your ass, William Blake. Rise now, and drive your cart and plough over the bones of the dead!”
;)
Upvotes: 0
Views: 50
Reputation: 14269
Try something like this:
WITH status AS (
SELECT DISTINCT client_id,
first_value(statement_status) OVER w1 AS last_status,
nth_value(statement_status, 2) OVER w1 AS prev_status
FROM records
WINDOW w1 AS (PARTITION BY client_id ORDER BY contact_ts DESC RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING)
)
SELECT CURRENT_DATE(),
SUM(last_status = 'interested') AS interesed,
SUM(last_status = 'not_interested' AND prev_status = 'not_present') AS not_interested
FROM status
Upvotes: 1