SQL detect change in row - sql

Question

I have data from sql server attached :
select * from log
What I want to do is I want to check if there any changes in code for the column name. So if you see the data from table log, the code change 2 times (B02,B03).
What I want to do is I want to retrieve the row which is the first changes everytime the code change. In this sample, the first changes is on the red box. So I want to have the result for row 5 and row 9.
I've tried to use partition like code below:
select a.name,a.code from(
select name,code,row_number() over(partition by code order by name) as rank from log)a
where a.rank=1
and get result like this.
However, I don't want the first row to be retrieved. Since it is the first value and I don't need that. So i just want to retrieve the changes indicates by column code. Please help if you know how to do it.
and please note, I can't write query using filter where code <> 'B01', because in this case, I don't know what is the first value.
Please assume the first value is the data that first inserted into the table.

Answer 1

Use lag to get the previous row's value (assuming id specifies ordering) and get the rows where it is different from the current row's value.
create table #log (id int identity(1,1) not null, name nvarchar(100), code nvarchar(100));
insert into #log(name,code) values ('SARUMA','B01'), ('SARUMA','B01'), ('SARUMA','B01'), ('SARUMA','B01');
insert into #log(name,code) values ('SARUMA','B02'), ('SARUMA','B02'), ('SARUMA','B02'), ('SARUMA','B02');
insert into #log(name,code) values ('SARUMA','B03'), ('SARUMA','B03');
select name
,code
from (
select l.*
,lag(code) over (
partition by name order by id
) as prev_code
from #log l
) l
where prev_code <> code

Answer 2

create table #log (name nvarchar(100), code nvarchar(100));
insert into #log values ('SARUMA','B01'), ('SARUMA','B01'), ('SARUMA','B01'), ('SARUMA','B01');
insert into #log values ('SARUMA','B02'), ('SARUMA','B02'), ('SARUMA','B02'), ('SARUMA','B02');
insert into #log values ('SARUMA','B03'), ('SARUMA','B03');
-- remove duplicates
with Singles (name, code)
AS (
select distinct name, code from #log
),
-- At first you need an order, in time? By alphanumerical code? Otherwise you cannot decide which is the first item you want to remove
-- So I added an identity ordering, but it is preferable to use a physical column
OrderedSingles (name, code, id)
AS (
select *, row_number() over(order by name)
from Singles
)
-- Now self-join to get the next one, if the index is sequential you can join id = id+1
-- and take the join columns
select distinct ii.name, ii.Code
from OrderedSingles i
inner join OrderedSingles ii
on i.Name = ii.Name and i.Code <> ii.Code
where i.id < ii.Id;

Answer 3

I think that your original post was pretty close, though you would want the windowing function to be on the [NAME] column, not the code. Please see my modifications, below. I've also changed the predicate to be >1, as 1 would be the original record.
SELECT
a.[name]
,a.[code]
FROM (
SELECT
[name]
,[code]
,ROW_NUMBER() OVER(PARTITION BY [name] order by [name], [code]) AS [rank]
FROM log)a
WHERE a.rank>1
NOTE: you may want to not use NAME as a field, since it is a reserved word. Additionally, RANK is a reserved word as well, and you've used it to alias the ROW_NUMBER in the nested query. You may want to use another non-reserved word for that - personally, I use RANKED for that purpose.