Question

Here is a sample table that mimics my scenario:

COL_1   COL_2       
1       a               
2       a               
9       b               
8       b               
11      b               

I have two columns where Column 1 have different numerical values and Column 2 represents different classes (a, b, c, etc.). In column 3 I would like to have them counted per class and matched with 0/1 numbers based on what was the biggest/lowest value in Column 1. So if in my example class “a” have only two values (in our example 1 and 2) they should be counted from 0 to 1 (where 0 stands for lowest number and 1 for biggest from Column 1). If class b is being represented by 3 entries I would like to have them matched by 0/1/2 numbers based on what is the biggest/lowest number from Column 1 in “b” class.

After my query I woul like to have something like this:

COL_1   COL_2   COL_3   
1       a       0       
2       a       1       
9       b       1       
8       b       0       
11      b       2       

So, in other words I want a query that looks how many duplicates we have in column 2 and gives them values from lowest to biggest (where 0 is lowest) based on what they are represented in column 1.

Thanks to anyone who helps me with this.

*I am using MS SQL, if that would make any difference.

Answer 1

select 
    col_1,
    col_2,
    row_number() over(
        partition by col_2 order by col_1
    ) - 1 as col3
from yourtablename