Timestamp & data point search

Question

I should preface this by saying I only have read only access to this Oracle SQL database and I'm using TOAD.
looking for a query to find data points where two types of data points exist
So I have 4 columns: STID(id of device), CASE_SAK(unique id), timestamp(in form MM/DD/YY HH:MI:SS AM or PM), and COMMAND_ID.
COMMAND_ID has three types of data in it: EOL_READ, ODO_READ, and PREDICTION
Any time there is a reading there are 1 of 4 possibilities occur.
I can get EOL/ODO/PREDICTION, EOL/ODO, or EOL, or ODO
example:
COLUMNS: STID, CASE_SAK, timestamp, COMMAND_ID
ROW # 1: 2545, 100, 8/20/2006 12:00:00 AM, EOL_READ
ROW # 2: 2545, 100, 8/20/2006 12:00:00 AM, ODO_READ
ROW # 3: 2545, 100, 8/20/2006 12:00:00 AM, PREDICTION
ROW # 4: 3125, 101, 8/21/2006 12:01:00 AM, EOL_READ
ROW # 5: 3125, 101, 8/21/2006 12:01:00 AM, ODO_READ
Now I would like to query where exists a timestamp for each month in the last 6 months because these are monthly readings and I need to give only the data where there is a reading in each of the six months(no nulls) from the beginning of this month and going back 6 months, and I also need to only pull those cases where exists both an EOL_READ and an ODO_READ at the very least, but I also need to pull a PREDICTION as well where it exists. I am scratching my head on this one.
In addition I need to pull at first only 1000 unique STIDs, and then I need to plug in the 1000 STIDs into the same query for the future.
I realize this is a lot to ask the board, but any help or ideas would be greatly appreciated.

Answer 1

Not sure if this fully meets your requirements but it should be close;
create table query_results(stid number, case_sak number, time_stamp timestamp, command_id varchar2(20));
with test_data as (  -- middle 6 months of data qualifies 
   select 2545 stid, 100 case_sak, to_timestamp('07/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM') time_stamp, 'EOL_READ' command_id
      from dual union all
   select 2545, 100, to_timestamp('08/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2545, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2545, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2545, 100, to_timestamp('10/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'PREDICTION' from dual union all
   select 2545, 101, to_timestamp('11/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'EOL_READ' from dual union all
   select 2545, 101, to_timestamp('12/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2545, 101, to_timestamp('01/21/2007 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2545, 101, to_timestamp('02/21/2007 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   -- 6 months of data, no EOL_READ 
   select 3125, 100, to_timestamp('08/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 3125, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 3125, 100, to_timestamp('10/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'PREDICTION' from dual union all
   select 3125, 101, to_timestamp('11/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'PREDICTION' from dual union all
   select 3125, 101, to_timestamp('12/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 3125, 101, to_timestamp('01/21/2007 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   -- 6 months of data with gap 
   select 1111, 100, to_timestamp('07/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'EOL_READ' from dual union all
   select 1111, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 1111, 100, to_timestamp('10/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'PREDICTION' from dual union all
   select 1111, 101, to_timestamp('11/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'EOL_READ' from dual union all
   select 1111, 101, to_timestamp('12/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 1111, 101, to_timestamp('01/21/2007 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   -- middle 6 months of data qualifies 
   select 2222, 100, to_timestamp('07/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 100, to_timestamp('08/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 100, to_timestamp('09/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 100, to_timestamp('10/20/2006 12:00:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'PREDICTION' from dual union all
   select 2222, 101, to_timestamp('11/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'EOL_READ' from dual union all
   select 2222, 101, to_timestamp('12/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 101, to_timestamp('12/21/2006 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 101, to_timestamp('01/21/2007 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual union all
   select 2222, 101, to_timestamp('02/21/2007 12:01:00 AM','MM/DD/YYYY HH:MI:SS AM'), 'ODO_READ' from dual)
select *
from (
   select
      t.stid,
      t.case_sak,
      t.time_stamp,
      t.command_id,
      count(distinct trunc(t.time_stamp,'mm')) over (partition by stid) month_count,
      count(case when command_id='EOL_READ' then 1 else NULL end) over (partition by stid) eol_read_count,
      count(case when command_id='ODO_READ' then 1 else NULL end) over (partition by stid) odo_read_count
   from test_data t,
      (select trunc(add_months(sysdate, -1 * rownum),'mm') month_id
      from dual
      connect by rownum <= 6) d
   where d.month_id = trunc(t.time_stamp,'mm')
   order by stid, time_stamp, case_sak, command_id
) sub
where month_count = 6
and eol_read_count > 0
and odo_read_count > 0
and rownum < = 1000
and not exists (select 1
                from query_results r
                where sub.stid = r.stid
                and sub.case_sak = r.case_sak
                and sub.time_stamp = r.time_stamp)
order by stid, time_stamp, case_sak, command_id;You may want to make a view out of the query and use it to populate a table that retains the 1000 records that you need. This query assumes that the historical records are stored in a table named query_results. You can drop the test_data selects and replace the table reference in the query with your table. You can also remove the three count columns from the SELECT.
I recommend that you rename the 'timestamp' column as this is an Oracle reserved word and it has caused problems for others in the past.
Message was edited by:
MScallion
Added distinct