Selecting multiple unique matches from one column that match another column - sql

Question

I have a list of codes (101, 102, 103, 104) and I want to pick out the people in the following table that have two or more different codes from the list occurring within a year of each other.
Name Code1 Code1date
John 101 01/01/2016
John 102 01/02/2013
Chris 101 01/01/2015
Chris 101 01/05/2014
Chris 102 01/10/2015
Mark 101 01/11/2011
Mark 101 01/01/2011
Mark 107 01/07/2012
So in this sample only Chris would be selected because he has a 101 code and a 102 code within a year of each other.
Thanks!

Answer 1

Try with the below query if you are using SQL Server.
SELECT name
FROM yourtable
WHERE code1 in (101, 102, 103, 104)
GROUP BY name, year(code1date)
HAVING COUNT(distinct code1) > 1

Answer 2

Version 1: Same year: happened in the same year
You need to make a group for each name and year and only show those distinct names that have more than 1 unique code:
select distinct name
from sample_table
where code1 between 101 and 104
group by name, extract(year from code1date)
having count(distinct code1) > 1
This will result in Chris only being presented in output.
EXTRACT function is ANSI-SQL compliant, but it will work assuming that code1date is of date type
In case it is of text data type, you could get 4 characters from the right, so for example right(code1date, 4)
Version 2: Same year: scan back- and onwards for one year difference
If by one year you mean not the same year, but scanning backwards and onwards from a date for 1 year difference, then here's the solution for Postgres:
SELECT
a.name
FROM sample_table a
JOIN sample_table b ON
a.name = b.name
AND a.code1 <> b.code1
AND b.code1date BETWEEN a.code1date - interval '1 year' AND a.code1date + interval '1 year'
WHERE a.code1 BETWEEN 101 AND 104
GROUP BY a.name
HAVING COUNT(DISTINCT a.code1) > 1
Above also assumes that your code1date is of date type. If that's not the case then you should think about converting it to a proper format. If that's beyond your reach, then you could always get the last character from your column, cast to Integer, increment it and append it back to the substring without the last char thus replacing the value of year :-)

Answer 3

SELECT
t1.Name
FROM
TableName t1
INNER JOIN TableName t2
ON t1.Name = t2.Name
AND t2.Code1Date BETWEEN DATEADD(year,-1,t1.Code1Date) AND DATEADD(year,1,t1.Code1Date)
AND t1.Code1 <> t2.Code1
AND t2.Code1 IN (101,102,103,104)
WHERE
t1.Code1 IN (101,102,103,104)
GROUP BY
t1.Name
HAVING
COUNT(DISTINCT t1.Code1) > 1
So this is for sql-server but just change up the BETWEEN statement to reflect the date functions of what ever rdbms you are working with.