Counting number of sequential occurences for a person - SQL

Question

I have a table like:

Name | ID | Event
Smith| 1  | 
Smith| 2  | Y
Smith| 3  | 
Jones| 1  | 
Jones| 2  | Y
Jones| 3  | 
Jones| 4  | Y

I'd like to count the number of times an Event has been seen for each person at each point, e.g.:

Name | ID | Event | Event Count
Smith| 1  |       | 0
Smith| 2  | Y     | 1
Smith| 3  |       | 1
Jones| 1  |       | 0
Jones| 2  | Y     | 1
Jones| 3  |       | 1
Jones| 4  | Y     | 2

I'm guessing I can't do this in SQL? If not, can you be very clear how I go about doing this in SAS (or whatever way is appropriate), as I am new to this!

(FYI, this is leading to me being able to differentiate rows that happen before or after each event - i.e. filter by Event = blank, and anything 0 happened before the first event, anything 1 after, etc. There might be an easier way to do this.)

Thanks!

Answer 1

If you want to go down the SAS route, it reads data sequentially so is very good at this type of problem

data have;
infile datalines missover;
input Name $ ID  Event $;
datalines;
Smith 1   
Smith 2   Y
Smith 3   
Jones 1   
Jones 2   Y
Jones 3   
Jones 4   Y
;
run;

proc sort data=have;
by name id;
run;

data want;
set have;
by name id;
if first.name then event_count=0;
event_count+(event='Y');
run;

Answer 2

You could potentially do something like this in a query:

select Name, ID, Event,
    (
        select count(*)
        from MyTable
        where Name = t.Name
            and Event = 'Y'
            and ID <= t.ID
    ) as EventCount
from MyTable t

The correlated subquery will find this count for you, though this is something of a triangular join (SQL Server link, but still applicable), so performance isn't wonderful.

Here is the SQL Fiddle showing the result.

Note that this should work in virtually any RDBMS.

Answer 3

SELECT Name, ID, Event, grpTotal
FROM
  (
    select  Name,
            ID,
            Event,
            @sum := if(@grp = Name,@sum,0) + if(`Event` = 'Y',1,0) as grpTotal,
            @grp := Name
    from    TableName,
            (select @grp := '', @sum := 0) vars
    order   by  Name, ID
  ) s

• SQLFiddle Demo