Question

I have an SQL table like that without row_number column,

acct_id     part_id   sub_model   row_number
111         201301      A            1
111         201302      A            2
111         201303      B            1
111         201304      B            2 
111         201305      C            1
111         201306      C            2
111         201307      C            3
111         201308      A            1
111         201309      A            2
111         201310      A            3
111         201311      A            4
222         201301      B            1
222         201302      B            2
222         201303      C            1
222         201304      C            2

and so on. I want to add row_number column whenever a model has changed for the Acct_ID, row number will start from 1 and count till there is a change in model. Row_numbers of different acct_IDs shouldn't effect each other. So row_number should count and start again for a specific ID. If next acct_ID starts with same model with last row of previous acct_ID, it should notice and start from 1 again. This is easy to do in excel, at least for me. But data is over a million row. So I have to handle it with SQL and no idea how to do it.

Answer 1

This would be my suggestion (sql fiddle):

SELECT T1.*,
   T2.row_number
FROM Table1 AS T1
JOIN (
SELECT  acct_id,
            part_id,
            sub_model,
            COUNT(*) AS row_number
FROM Table1
GROUP BY acct_id, sub_model) AS T2 ON T2.acct_id = T1.acct_id AND T2.sub_model = T1.sub_model
ORDER BY T1.acct_id, T1.part_id, T1.sub_model

I'm summing up the amount of times on sub_model has been used on a acct_id. I have no idea whether this information is actually what you are looking for, if not i think Mike Brant is correct in saying this is better done in the view layer in php or whatever language you are using.

Counting duplicated Data on SQL

1 个答案: