Using CTE to count number of rows in inner query - sql

Question

I'm learning CTE and I've encounter an exercise on I cannot solve. It is not a homework, but an exercise from an online course I've taken to learn SQL. I'm interested in where I've made a mistake and some explanation so answering with only the correct code will not help me to learn CTE.
The task is to count projects that raised 100% to 150% of the minimum amount, and those that raised more than 150%.
I've written the following CTE:
WITH nice_proj AS
(SELECT project_id AS pid,
amount AS amount,
minimal_amount AS minimal
FROM donation d
INNER JOIN project p ON (d.project_id = p.id)
GROUP BY pid,
minimal,
amount
HAVING sum(amount) >= minimal_amount)
SELECT count(*) AS COUNT,
(CASE
WHEN sum(amount)/minimal <=1.5 THEN 'good projects'
ELSE 'great projects'
END) AS tag
FROM nice_proj
GROUP BY minimal;
The query returns nothing but it should produce something similar to:
+-------+----------------+
| count | tag |
+-------+----------------+
| 16 | good projects |
+-------+----------------+
| 7 | great projects |
+-------+----------------+
Please have a look at the tables (they are truncated):
donation
+----+------------+--------------+---------+------------+------------+
| id | project_id | supporter_id | amount | amount_eur | donated |
+----+------------+--------------+---------+------------+------------+
| 1 | 4 | 4 | 928.40 | 807.70 | 2016-09-07 |
+----+------------+--------------+---------+------------+------------+
| 2 | 8 | 18 | 384.38 | 334.41 | 2016-12-16 |
+----+------------+--------------+---------+------------+------------+
| 3 | 6 | 12 | 367.21 | 319.47 | 2016-01-21 |
+----+------------+--------------+---------+------------+------------+
| 4 | 2 | 19 | 108.62 | 94.50 | 2016-12-29 |
+----+------------+--------------+---------+------------+------------+
| 5 | 10 | 20 | 842.58 | 733.05 | 2016-11-30 |
+----+------------+--------------+---------+------------+------------+
| 6 | 4 | 15 | 653.76 | 568.77 | 2016-08-05 |
+----+------------+--------------+---------+------------+------------+
| 7 | 4 | 14 | 746.52 | 649.48 | 2016-08-03 |
+----+------------+--------------+---------+------------+------------+
| 8 | 10 | 3 | 962.36 | 837.25 | 2016-10-30 |
+----+------------+--------------+---------+------------+------------+
| 9 | 1 | 20 | 764.05 | 664.72 | 2016-08-24 |
+----+------------+--------------+---------+------------+------------+
| 10 | 10 | 4 | 1033.42 | 899.08 | 2016-02-26 |
+----+------------+--------------+---------+------------+------------+
| 11 | 5 | 6 | 571.90 | 497.55 | 2016-10-06 |
+----+------------+--------------+---------+------------+------------+
project
+----+------------+-----------+----------------+
| id | category | author_id | minimal_amount |
+----+------------+-----------+----------------+
| 1 | music | 1 | 1677 |
+----+------------+-----------+----------------+
| 2 | music | 5 | 21573 |
+----+------------+-----------+----------------+
| 3 | travelling | 2 | 4952 |
+----+------------+-----------+----------------+
| 4 | travelling | 5 | 3135 |
+----+------------+-----------+----------------+
| 5 | travelling | 2 | 8555 |
+----+------------+-----------+----------------+
| 6 | video | 4 | 6835 |
+----+------------+-----------+----------------+
| 7 | video | 4 | 7978 |
+----+------------+-----------+----------------+
| 8 | games | 1 | 4560 |
+----+------------+-----------+----------------+
| 9 | games | 2 | 4259 |
+----+------------+-----------+----------------+
| 10 | games | 1 | 5253 |
+----+------------+-----------+----------------+

Answer 1

My advice is to aggregate the donations table first, then compare it to the project table.
By doing this the join between donations and project is always 1:1. This in turn means you avoid having to group by "values" (minimal_amount), instead only grouping by "identifiers" (project_id).
WITH
donation_summary AS
(
SELECT
project_id,
SUM(amount) AS total_amount
FROM
donation
GROUP BY
project_id
)
SELECT
CASE WHEN d.total_amount <= p.minimal_amount * 1.5
THEN 'good projects'
ELSE 'great projects'
END
AS tag,
COUNT(*) AS project_count
FROM
donation_summary AS d
INNER JOIN
project AS p
ON p.id = d.project_id
WHERE
d.total_amount >= p.minimal_amount
GROUP BY
tag
That said, I'd normally use the following final query and get two columns rather than two rows...
SELECT
SUM(CASE WHEN d.total_amount <= p.minimal_amount * 1.5 THEN 1 ELSE 0 END) AS good_projects,
SUM(CASE WHEN d.total_amount > p.minimal_amount * 1.5 THEN 1 ELSE 0 END) AS great_projects
FROM
donation_summary AS d
INNER JOIN
project AS p
ON p.id = d.project_id
WHERE
d.total_amount >= p.minimal_amount

Answer 2

You need to remove amount from the grouping, this should return the expected result:
WITH nice_proj AS
(SELECT project_id AS pid,
sum(amount) AS amount,
minimal_amount AS minimal
FROM donation d
INNER JOIN project p ON (d.project_id = p.id)
GROUP BY pid,
minimal
HAVING sum(amount) >= minimal_amount)
SELECT count(*) AS COUNT,
(CASE
WHEN amount/minimal <=1.5 THEN 'good projects'
ELSE 'great projects'
END) AS tag
FROM nice_proj
GROUP BY tag;