Count distinct and non distinct values

Question

I have an sql table looking like this:

+----+-----------+------------+
| Id | EnquiryId | PropertyId |
+----+-----------+------------+
|  1 |         1 |        20  |
|  2 |         1 |        25  |
|  3 |         1 |        26  |
|  4 |         2 |        20  |
|  5 |         3 |        20  |
|  6 |         4 |        20  |
+----+-----------+------------+

I want to count how many enquiries propertyid 20 has on it's own, and how many that is shared with other properties

So the result should be something like this:

Number of single enguiry: 3

Number of shared enquiries: 1

It's perfectly fine if it requires two select statements :)

The attempts so far looks like this:

(select count(distinct [EnquiryId]) 
from [EnquiryProperty] where propertyid=20) as 'SingleEnquiry'

This gave me 4 results (I need it to be 3)

(select count([PropertyId]) FROM [EnquiryProperty] where propertyid=20 GROUP BY propertyid HAVING COUNT(*) > 1) as 'MultipleEnquiry'

And this gave me 0 results

My attemts so far looks like this: ,(select count(distinct [EnquiryId]) from [EnquiryProperty] where propertyid=325) as 'SingleEnquiry' ,(select count([PropertyId]) FROM [EnquiryProperty] where propertyid=325 GROUP BY propertyid HAVING COUNT(*) > 1) as 'MultipleEnquiry' — Marcus Ohlsson
– Marcus Ohlsson, Commented Oct 19, 2017 at 10:19

Srini V · Accepted Answer · 2017-10-19 10:27:23Z

5

One method is two levels of aggregation. The inner level assigns flags to each enquiry. The second uses those to get the information you want:

select sum(is_20 * (1 - is_not_20)) as single_enquiry,
       sum(is_20 * is_not_20) as shared_enquiry
from (select enquiryid,
             max(case when propertyid = 20 then 1 else 0 end) as is_20,
             max(case when propertyid <> 20 then 1 else 0 end) as is_not_20
      from t
      group by enquiryid
     ) e;

edited Oct 19, 2017 at 10:27

Srini V

11.5k15 gold badges70 silver badges93 bronze badges

answered Oct 19, 2017 at 10:25

Gordon Linoff

1.3m62 gold badges705 silver badges857 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Marcus Ohlsson Over a year ago

Thank you verry much @SriniV, this works like a sharm :)

CodeSmith Over a year ago

I like the idea with flagging 1st then aggregating the flags. Gonna keep that one in memory :)

Srini V Over a year ago

@MarcusOhlsson, it was actually gordonlinoff to be thanked

CodeSmith · Accepted Answer · 2017-10-19 10:34:14Z

1

You can do same in 2 steps (its a tad easier to understand and has no nesting):

select count(PropertyId) AS Single20s
from WhateverTableIsCalled
group by EnguryId
having count(PropertyId) = 1

select count(PropertyId) AS Shared20s
from WhateverTableIsCalled
group by EnguryId
having count(PropertyId) > 1

answered Oct 19, 2017 at 10:34

CodeSmith

3,2171 gold badge19 silver badges34 bronze badges

Collectives™ on Stack Overflow

Count distinct and non distinct values

2 Answers 2

3 Comments

Comments

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Linked

Related