mysql rank and subtraction of count - mysql

I have a table to store the votes. I query out the rank of candidates, and I also want the candidate to see how many votes are required to equal the votes held by the candidate ranked immediately above them.
CREATE TABLE `vote` (
`id` bigint(20) NOT NULL AUTO_INCREMENT,
`candidateid` int(11) NOT NULL,
`openid` varchar(2048) NOT NULL,
`weight` int(11) DEFAULT '1',
`time` bigint(20) DEFAULT NULL,
`date` varchar(56) DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=73 DEFAULT CHARSET=utf8;
select t.* , #curRank := #curRank + 1 AS rank
from
(
SELECT candidateid,
count(*) as num
FROM vote p
group by candidateid
ORDER BY num desc
) t, (SELECT #curRank := 0) r
As far as I got
candidateid num rank
1 42 1
6 16 2
8 9 3
2 3 4
7 1 5
4 1 6
I want to get
candidateid num sub rank
1 42 0 1
6 16 26 2
8 9 7 3
2 3 6 4
7 1 2 5
4 1 0 6
e.g. candidateid=6 requires 26 votes to equal the candidate ranked above them. candidateid=2 only needs 6 votes to reach 9, drawing level with candidateid=8.


Just extend your query with an additional variable to calculate the difference:
select t.candidateid , #curRank := #curRank + 1 AS rank, if(#prevote=-1, 0,#prevote-t.num) as sub, #prevote:=t.num as num
from
(
SELECT candidateid,
count(*) as num
FROM vote p
group by candidateid
ORDER BY num desc
) t, (SELECT #curRank := 0, #prevote:=-1) r

Related

MySQL: Select newest two rows per Group

I have a table like this:
CREATE TABLE `data` (
`id` int(11) NOT NULL,
`deviceId` int(11) NOT NULL,
`position_x` int(11) NOT NULL,
`position_y` int(11) NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;
ALTER TABLE `data`
ADD PRIMARY KEY (`id`);
COMMIT;
id, deviceId, position_x, position_y
1 1 100 0
2 2 150 50
3 3 200 20
4 1 220 20
5 1 210 10
6 2 100 40
7 3 120 50
8 3 130 60
9 2 240 15
I need the "newest" two rows per DeviceID, where a bigger ID means newer.
Right now, I'm selecting the newest row per Device via this query:
SELECT
id,
deviceId,
position_x, position_y
FROM data
WHERE deviceId > 0 AND
id IN (SELECT MAX(id) FROM data GROUP BY deviceId)
And in a loop, where I output the data, I select the second latest row for every deviceId in an individual query, which is kinda slow/dirty:
SELECT
position_x
position_y
FROM data
WHERE deviceId = :deviceId AND
id < :id
ORDER BY id DESC
LIMIT 1
Is there a way to combine both queries or at least, in one query, select the second row for every deviceId from query 1?
Thanks

You can try using row_number()
select * from
(
SELECT
id,
deviceId,
position_x, position_y,row_number() over(partition by deviceid order by id desc) as rn
FROM data
WHERE deviceId > 0
)A where rn=2

You can use a correlated subquery for this as well:
SELECT d.*
FROM data d
WHERE d.deviceId > 0 AND
d.id = (SELECT d2.id
FROM data d2
WHERE d2.deviceId = d.deviceId
ORDER BY d2.id DESC
LIMIT 1, 1
);
With an index on data(deviceId, id desc), you might be impressed at the performance.

Aggregate rows by id comparing column values

I have the following table that groups users by their permissions
userIds permissions
4,5,7,8 100,1600,500,501,502,400,401,1500,1501
The numbers in the permissions column are the sections ids.
Some of these sections may have other data associated which I retrieved and stored in another table.
sectionId userId resourceId
100 4 NULL
1600 4 NULL
500 4 NULL
501 4 NULL
502 4 NULL
400 4 NULL
401 4 1
1500 4 NULL
1501 4 NULL
100 5 NULL
1600 5 NULL
500 5 NULL
501 5 NULL
502 5 NULL
400 5 NULL
401 5 1,2
1500 5 NULL
1501 5 NULL
100 7 NULL
1600 7 NULL
500 7 NULL
501 7 NULL
502 7 NULL
400 7 NULL
401 7 2
1500 7 NULL
1501 7 NULL
100 8 NULL
1600 8 NULL
500 8 NULL
501 8 NULL
502 8 NULL
400 8 NULL
401 8 1
1500 8 NULL
1501 8 NULL
My goal is to compare, for each user in the userIds column of the first table (splitted by comma), every row of the second table in order to check if each user has the same resourceId value for that specific sectionId.
If one or more users have the same resourceId value for each section I want to keep them group together, otherwise they need to be on different rows.
This is the output I'm expecting from the sample data provided:
userIds permissions
4,8 100,1600,500,501,502,400,401,1500,1501
5 100,1600,500,501,502,400,401,1500,1501
7 100,1600,500,501,502,400,401,1500,1501
UPDATE
I managed to get the desidered output in the following way:
-- Numbers table creation
DROP temporary TABLE IF EXISTS tally;
CREATE temporary TABLE tally
(
n INT NOT NULL auto_increment PRIMARY KEY
);
INSERT INTO tally
(n)
SELECT NULL
FROM (SELECT 0 AS N
UNION ALL
SELECT 1
UNION ALL
SELECT 2
UNION ALL
SELECT 3
UNION ALL
SELECT 4
UNION ALL
SELECT 5
UNION ALL
SELECT 6
UNION ALL
SELECT 7
UNION ALL
SELECT 8
UNION ALL
SELECT 9) a,
(SELECT 0 AS N
UNION ALL
SELECT 1
UNION ALL
SELECT 2
UNION ALL
SELECT 3
UNION ALL
SELECT 4
UNION ALL
SELECT 5
UNION ALL
SELECT 6
UNION ALL
SELECT 7
UNION ALL
SELECT 8
UNION ALL
SELECT 9) b;
-- Split users by comma from first table
DROP temporary TABLE IF EXISTS tmppermissions2;
CREATE temporary TABLE tmppermissions2
(
userid VARCHAR(255) NOT NULL,
permissions TEXT NOT NULL
);
INSERT INTO tmppermissions2
SELECT userid,
permissions
FROM (SELECT Substring_index(Substring_index(t.userids, ',', tally.n), ',', -1
)
userId,
t.permissions
permissions
FROM tally
INNER JOIN tmppermissions t
ON Char_length(t.userids) - Char_length(
REPLACE(t.userids, ',',
'')) >=
tally.n - 1
ORDER BY n) AS split;
-- Gets the users with the same permissions
DROP temporary TABLE IF EXISTS sharedprofiles;
CREATE temporary TABLE sharedprofiles
(
userids VARCHAR(255) NOT NULL,
permissions TEXT NOT NULL,
profileid INT(11)
);
INSERT INTO sharedprofiles
SELECT Group_concat(userid),
permissions,
NULL
FROM tmppermissions2
WHERE userid NOT IN (SELECT split.userid
FROM (SELECT Substring_index(Substring_index(r.userids,
',',
t.n), ',', -1)
userId
FROM tally t
INNER JOIN tmppermissions r
ON Char_length(r.userids)
- Char_length(
REPLACE(r.userids, ',',
'')) >=
t.n - 1
WHERE Position(',' IN r.userids) > 0
ORDER BY n) AS split
WHERE split.userid IN (SELECT *
FROM (SELECT Group_concat(userid
ORDER
BY userid ASC)
AS
users
FROM
tmpcurrentresources2
GROUP BY resourceid,
sectionid
ORDER BY users) b
WHERE Position(',' IN b.users) =
0))
GROUP BY permissions
ORDER BY Group_concat(userid);
-- Gets the users with specific permissions
DROP temporary TABLE IF EXISTS singleprofiles;
CREATE temporary TABLE singleprofiles
(
userid VARCHAR(255) NOT NULL,
permissions TEXT NOT NULL,
profileid INT(11)
);
INSERT INTO singleprofiles
SELECT userid,
permissions,
NULL
FROM tmppermissions2
WHERE userid IN (SELECT split.userid
FROM (SELECT Substring_index(Substring_index(r.userids, ',',
t.n),
',', -1)
userId
FROM tally t
INNER JOIN tmppermissions r
ON Char_length(r.userids) -
Char_length(
REPLACE(r.userids, ',',
'')) >=
t.n - 1
WHERE Position(',' IN r.userids) > 0
ORDER BY n) AS split
WHERE split.userid IN (SELECT *
FROM (SELECT Group_concat(userid
ORDER BY
userid ASC)
AS
users
FROM tmpcurrentresources2
GROUP BY resourceid,
sectionid
ORDER BY users) b
WHERE Position(',' IN b.users) = 0))
ORDER BY userid;
-- Merge the results
SELECT *
FROM sharedprofiles
UNION
SELECT *
FROM singleprofiles;
I'm wondering if there is a more concise way to accomplish the same result.

The solution (as I suspect you already know) is to normalise your schema.
So instead of...
userIds permissions
4,5 100,1600,500
...you might have
userIds permissions
4 100
4 1600
4 500
5 100
5 1600
5 500

Get the Top records from each group in MYSQL

I've a table event_log with the following columns in MYSQL,
CREATE TABLE IF NOT EXISTS `event_log` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`customer_id` varchar(50) DEFAULT NULL,
`event_time` datetime DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1 AUTO_INCREMENT=1 ;
The sample Data can be,
id, customer_id, event_time
1 100 '2015-03-22 23:54:37'
2 100 '2015-03-21 23:54:37'
3 100 '2015-03-20 23:54:37'
4 101 '2015-03-19 23:54:37'
5 102 '2015-03-19 23:54:37'
6 102 '2015-03-18 23:54:37'
7 103 '2015-03-17 23:54:37'
8 103 '2015-03-16 23:54:37'
9 103 '2015-03-15 23:54:37'
10 103 '2015-03-14 23:54:37'
I want to group on customer_id and then pick the top 2 records from each group using event_time column (whose time is greater)
Please, suggest
Thanks,
Faisal Nasir

Here is a version that doesn't use variables:
select el.*
from event_log el
where 2 >= (select count(*)
from event_log el2
where el2.customer_id = el.customer_id and
el2.event_time >= el.event_time
);
This could even have reasonable performance with an index on event_log(customer_id, event_time).

One way to use user defined variables to pick 2 recent entries per customer_id
SELECT `id`, `customer_id`, `event_time`,row_num
FROM (
SELECT *,
#r:= CASE WHEN #g = `customer_id` THEN #r +1 ELSE 1 END row_num,
#g:= `customer_id`
FROM event_log
CROSS JOIN(SELECT #g:= NULL,#r:=0) a
ORDER BY `customer_id`,`event_time` desc
) t
where row_num <= 2
DEMO

How to rank MySQL results, based on different values?

I have 2 different tables in my database by the name of: rank, settings.
Here is how each table looks like with a few records in them:
Table #rank:
id points userid
-- ----- ------
1 500 1
2 300 2
3 900 3
4 1500 4
5 100 5
6 700 6
7 230 7
8 350 8
9 850 9
10 150 10
Table #settings:
userid active
------ ------
1 0
2 1
3 1
4 1
5 1
6 0
7 1
8 1
9 0
10 1
What I basically want to achieve is to select a specific row from #rank by ID, sort it by points and select 3 rows above the specific ID and 3 row below the specific ID but only for rows where the active column (from #settings) for the user equals 1.
For example:
I would like to select from #rank the ID of 8, and it should return me the following:
rank points userid
---- ----- ------
2 150 10
3 230 7
4 300 2
5 350 8
6 900 3
7 1500 4
I have created quite an extensive query for this, but the problem is, that it is ranking the columns before it decides that the user is active or not. However I need to rank the columns after it is decided that the user is active or not.
SELECT sub2.sort, sub2.points, sub2.userid
FROM
(
SELECT #sort1 := #sort1 + 1 AS sort, puu.points, puu.userid
FROM rank as puu,
(SELECT #sort1 := 0) s
LEFT JOIN
(
settings as p11
)
ON puu.userid = p11.userid,
WHERE p11.active = 1
ORDER BY puu.points DESC
) sub1
INNER JOIN
(
SELECT #sort2:=#sort2+1 AS sort, p2.points, p2.userid
FROM rank as p2,
(SELECT #sort2 := 0) s
LEFT JOIN
(
settings as p12
)
ON p2.userid = p12.userid,
WHERE p12.active = 1
ORDER BY points DESC
) sub2
ON sub1.userid = :userid
AND sub2.sort BETWEEN (sub1.sort - 5) AND (sub1.sort + 5)
Can you guys find any solution for my problem? If you can provide an SQLfiddle demo, that would be really awesome!

SELECT sort, points, user_id, active FROM (
SELECT #pos := #pos + 1 AS sort, id, points, r.user_id, s.active,
IF(user_id = :userid, #userpos := #pos, 0)
FROM rank r
JOIN settings s USING(user_id)
JOIN (SELECT #pos := 0, #userpos := 0) p
WHERE s.active = 1
ORDER BY points DESC
) list
WHERE sort BETWEEN #userpos - 3 AND #userpos + 3
I made a fiddle here: sqlfiddle

Possibly using 2 unioned queries to get the values before the ranking:-
SELECT #rank:=#rank+1 AS rank, points, userid
FROM
(
SELECT id, points, userid
FROM
(
SELECT rank.id, rank.points, rank.userid
FROM rank
INNER JOIN
(
SELECT points
FROM rank
WHERE id = 8
) sub0
ON rank.points >= sub0.points
INNER JOIN settings
ON rank.userid = settings.userid
WHERE settings.active = 1
ORDER BY rank.points LIMIT 3
) sub1
UNION ALL
SELECT id, points, userid
FROM
(
SELECT rank.id, rank.points, rank.userid
FROM rank
INNER JOIN
(
SELECT points
FROM rank
WHERE id = 8
) sub0
ON rank.points < sub0.points
INNER JOIN settings
ON rank.userid = settings.userid
WHERE settings.active = 1
ORDER BY rank.points DESC LIMIT 3
) sub1
) sub2
CROSS JOIN (SELECT #rank:=0) sub3
ORDER BY points

MySQL- Select at least n rows per group

Suppose you have
id / value
1 2
1 3
1 6
2 3
3 1
3 3
3 6
And I want to retrieve at least n rows per id group, let's say n = 4. In addition, it would help if a counter is added as a column. So the results should be like:
counter / id / value
1 1 2
2 1 3
3 1 6
4 null null
1 2 3
2 null null
3 null null
4 null null
1 3 1
2 3 3
3 3 6
4 null null
regards

I'm assuming that the combination of id and value is unique. Here's how you can do it without using MySQL variables:
SELECT
a.n AS counter,
b.id,
b.value
FROM
(
SELECT
aa.n,
bb.id
FROM
(
SELECT 1 AS n UNION ALL
SELECT 2 AS n UNION ALL
SELECT 3 AS n UNION ALL
SELECT 4 AS n
) aa
CROSS JOIN
(
SELECT DISTINCT id
FROM tbl
) bb
) a
LEFT JOIN
(
SELECT aa.id, aa.value, COUNT(*) AS rank
FROM tbl aa
LEFT JOIN tbl bb ON aa.id = bb.id AND aa.value >= bb.value
GROUP BY aa.id, aa.value
) b ON a.id = b.id AND a.n = b.rank
ORDER BY
a.id,
a.n

The next blog post describes the solution to your query:
SQL: selecting top N records per group.
It requires an additional small table of numbers, which is utilized to "iterate" the top N values per group via String Walking technique.
It uses GROUP_CONCAT as a way to overcome the fact MySQL does not support Window Functions. This also means it's not a pretty sight!
An advantage of this technique is that it does not require subqueries, and can optimally utilize an index on the table.
To complete the answer to your question, we must add an additional columns: you have requested a counter per item per group.
Here's an example using the world sample database, choosing top 5 largest counties per continent:
CREATE TABLE `tinyint_asc` (
`value` tinyint(3) unsigned NOT NULL default '0',
PRIMARY KEY (value)
) ;
INSERT INTO `tinyint_asc` VALUES (0),(1),(2),(3),(4),(5),(6),(7),(8),(9),(10),(11),(12),(13),(14),(15),(16),(17),(18),(19),(20),(21),(22),(23),(24),(25),(26),(27),(28),(29),(30),(31),(32),(33),(34),(35),(36),(37),(38),(39),(40),(41),(42),(43),(44),(45),(46),(47),(48),(49),(50),(51),(52),(53),(54),(55),(56),(57),(58),(59),(60),(61),(62),(63),(64),(65),(66),(67),(68),(69),(70),(71),(72),(73),(74),(75),(76),(77),(78),(79),(80),(81),(82),(83),(84),(85),(86),(87),(88),(89),(90),(91),(92),(93),(94),(95),(96),(97),(98),(99),(100),(101),(102),(103),(104),(105),(106),(107),(108),(109),(110),(111),(112),(113),(114),(115),(116),(117),(118),(119),(120),(121),(122),(123),(124),(125),(126),(127),(128),(129),(130),(131),(132),(133),(134),(135),(136),(137),(138),(139),(140),(141),(142),(143),(144),(145),(146),(147),(148),(149),(150),(151),(152),(153),(154),(155),(156),(157),(158),(159),(160),(161),(162),(163),(164),(165),(166),(167),(168),(169),(170),(171),(172),(173),(174),(175),(176),(177),(178),(179),(180),(181),(182),(183),(184),(185),(186),(187),(188),(189),(190),(191),(192),(193),(194),(195),(196),(197),(198),(199),(200),(201),(202),(203),(204),(205),(206),(207),(208),(209),(210),(211),(212),(213),(214),(215),(216),(217),(218),(219),(220),(221),(222),(223),(224),(225),(226),(227),(228),(229),(230),(231),(232),(233),(234),(235),(236),(237),(238),(239),(240),(241),(242),(243),(244),(245),(246),(247),(248),(249),(250),(251),(252),(253),(254),(255);
SELECT
Continent,
SUBSTRING_INDEX(
SUBSTRING_INDEX(
GROUP_CONCAT(Name ORDER BY SurfaceArea DESC),
',', value),
',', -1)
AS Name,
CAST(
SUBSTRING_INDEX(
SUBSTRING_INDEX(
GROUP_CONCAT(SurfaceArea ORDER BY SurfaceArea DESC),
',', value),
',', -1)
AS DECIMAL(20,2)
) AS SurfaceArea,
CAST(
SUBSTRING_INDEX(
SUBSTRING_INDEX(
GROUP_CONCAT(Population ORDER BY SurfaceArea DESC),
',', value),
',', -1)
AS UNSIGNED
) AS Population,
tinyint_asc.value AS counter
FROM
Country, tinyint_asc
WHERE
tinyint_asc.value >= 1 AND tinyint_asc.value <= 5
GROUP BY
Continent, value
;