SQL-将结果列中的所有值相加，并使用同一查询中的另一列计算百分比

Question

I found it quite difficult to describe my problem in the title, but I will be more clear here. I have two tables users_friends and active_users ... here is the schema for both

CREATE TABLE `users_friends` (
  `id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `user_id` int(11) DEFAULT NULL,
  `friend_id` int(11) DEFAULT NULL,
  PRIMARY KEY (`id`),
  UNIQUE KEY `friend_id_user_id_index` (`friend_id`,`user_id`),
  KEY `user_id` (`user_id`)
) ENGINE=InnoDB AUTO_INCREMENT=7967354 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;


CREATE TABLE `active_users` (
  `id` int(10) unsigned NOT NULL AUTO_INCREMENT,
  `first` varchar(155) COLLATE utf8mb4_unicode_ci DEFAULT NULL,
) ENGINE=InnoDB AUTO_INCREMENT=150948970 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci;

I want to see the percentage of active_users who are following 1 person, 2 people, 3 people, etc, etc

I ran this query which gets me the number of users following X number of people

( SELECT count(t.total_follows) AS users_count, t.total_follows FROM ( ( SELECT count(*) total_follows, active_users.id FROM active_users INNER JOIN users_friends on users_friends.user_id = active_users.id GROUP BY active_users.id ) AS t ) GROUP BY t.total_follows ORDER BY t.total_follows )

Here are the results

users_count   total_follows
   5                   1
   3                   2
   2                   3

I can get the total sum of the users_count column by modifying the query above and summing up the users_count like this

select SUM(t1.users_count) as total_sum from ( < insert above query> ) as t1

but I don't know how to calculate the SUM of the users_count column divided by the users_count which would get me the desired result shown below

users_count    total_follows   percentage
   5                  1           50.0
   3                  2           30.0
   2                  3           20.0        

I know I can get this column value by doing this

(count(t.total_follows) / 10) * 100 as percentage

but I can't hardcode the total sum ( 10 ) in my query. I need the query to calculate everything in 1 run. How can I modify my query to achieve this ?

Answer 1

You might want to try using a "with" statement:

WITH x as
( 
SELECT count(t.total_follows) AS users_count, 
t.total_follows 
FROM ( ( SELECT count(*) total_follows, active_users.id 
FROM active_users 
INNER JOIN users_friends on users_friends.user_id = active_users.id 
GROUP BY active_users.id ) AS t ) 
GROUP BY t.total_follows 
ORDER BY t.total_follows )
)

SELECT *, COUNT/(SELECT SUM(users_count) FROM x)::FLOAT percentage from x

Answer 2

I ended up just computing the total sum of all the columns and storing them in a variable called @total_sum and then using that in a second query ... would have been nice if I could use the WITH statement like @Stoddard Meigs proposed, but my version of mysql doesn't support it

set @total_sum = (select SUM(t1.users_count) as total_sum from ( ( SELECT 
count(table1.total_follows) AS users_count, table1.total_follows FROM ( ( 
SELECT count(*) total_follows, active_users.id FROM active_users INNER JOIN 
users_friends on users_friends.user_id = active_users.id GROUP BY 
active_users.id ) AS table1 ) GROUP BY table1.total_follows ORDER BY 
table1.total_follows ) ) as t1);

( SELECT count(table1.total_follows) AS users_count, table1.total_follows, 
(count(table1.total_follows) / @total_sum) * 100 as percentage FROM ( ( SELECT 
count(*) total_follows, active_users.id FROM active_users INNER JOIN 
users_friends on users_friends.user_id = active_users.id GROUP BY 
active_users.id ) AS table1 ) GROUP BY table1.total_follows ORDER BY 
table1.total_follows )