Reputation: 450
I have fallowing SQL query (pseudo query):
SELECT
some columns [...]
COUNT(clicks.id) AS clicks,
COUNT(transactions.id) AS transactions
FROM
campaign
LEFT JOIN
clicks ON clicks.key = campaign.key
LEFT JOIN (
SELECT
id, key
FROM
transactions
GROUP BY
userkey
) transactions ON clicks.key = transactions.key;
GROUP BY
campaign.id
Query return good results. On example:
column | columns [..] | 34 | 10
column | columns [..] | 22 | 1
column | columns [..] | 34 | 17
So, records in table clicks they have transactions or a few transactions or they do not.
Haw Can I retun COUNT() clicks who have COUNT(transactions.id) = 0 and COUNT(transactions.id) > 0 ? On example:
column | columns [..] | 34 | 10 | 4 (count data from clicks table which have related transactions) | 30 (count data from clicks table which not have related transactions)
column | columns [..] | 22 | 1 | 6 | 16
column | columns [..] | 34 | 17 | 10 | 24
Tahnks for help.
@UPDATE:
I solved my problem adding second table. Now my SQL query looks like:
SELECT
some columns [...]
COUNT(clicks.id) AS clicks,
COUNT(transactions.id) AS transactions,
COUNT(clicks_count.id) as witchout_transactions,
(COUNT(clicks.id) - COUNT(clicks_count.id)) as witch_transactions
FROM
campaign
LEFT JOIN
clicks ON clicks.key = campaign.key
LEFT JOIN (
SELECT
id, key
FROM
transactions
GROUP BY
userkey
) transactions ON clicks.key = transactions.key
LEFT JOIN (
SELECT
clicks.id,
COUNT(transactions.id) AS transactions
FROM
clicks
LEFT JOIN transactions ON clicks.key = transactions.key
GROUP BY clicks.id
HAVING transactions = 0
) clicks_count ON clicks_count.id = clicks.id
GROUP BY
campaign.id
Upvotes: 1
Views: 1795
Reputation: 51973
What about adding a second join with the clicks table
JOIN (
SELECT id
FROM clicks
LEFT JOIN transactions ON clicks.key = transactions.key AND transactions.id != 0) clicks2 ON clicks.id = clicks2.id
And in the select clause use this to for the two columns
SELECT
some columns [...]
COUNT(clicks.id) AS clicks,
COUNT(transactions.id) AS transactions
COUNT(clicks2.id) as clicks_with
clicks - clicks_with AS clicks_without
Upvotes: 1
Reputation: 46229
If I understand correctly, you can try to use CASE WHEN
expression and COUNT
Because you didn' provide any sample data and expected result, so I can only provide pseudo-query.
SELECT...,
COUNT(CASE WHEN [have transactions condition] then 1 end),
COUNT(CASE WHEN [not have related transactions condition] then 1 end)
If that didn't help you, you can provide some data and expect result, I will edit my answer.
Upvotes: 2