SQL Server Filtering JOIN Data

Question

I have two tables for my process instrument data, First table has the Machine ID, Tripping Time, Tripping reason

Machine ID  Trip Time              Trip Reason
XA-065      03-20-2014 09:40:098    ANY
XA-065      03-24-2014 18:33:040    ANY
XA-765      03-23-2014 22:16:002    ANY
XA-070      03-21-2014 15:17:023    ANY

Second table has the Machine ID , Starting time.

Machine ID  Start Time             Operator
XA-065      03-20-2014 12:40:098    ANY
XA-065      03-24-2014 20:33:040    ANY
XA-765      03-23-2014 23:16:002    ANY
XA-070      03-21-2014 18:17:023    ANY

I’ve to join the two tables, So, for each machine I can get Machine ID,Trip Time, Start Time, and then I can add a calculated column to get the downtime "Start Time-Trip Time" The problem is, when the same machine starts and trips several times, JOIN operation matches all possible Trip/Start combinations. Which results in wrong downtime calculations. This is the result of My Join, notice what happens with Machine XA-065 :

Machine ID  Trip Time               Start Time              Downtime
XA-065      03-20-2014 09:40:098    03-20-2014 12:40:098    3 Hours
XA-065      03-20-2014 09:40:098    03-24-2014 20:33:040    **11 hours**
XA-065      03-24-2014 18:33:040    03-20-2014 12:40:098    **-6 Hours**
XA-065      03-24-2014 18:33:040    03-24-2014 20:33:040    2 Hours
XA-765      03-23-2014 22:16:002    03-23-2014 23:16:002    1 Hour
XA-070      03-21-2014 15:17:023    03-21-2014 18:17:023    3 Hours

Because JOIN will take all possible combinations for same Machine ID, I’m getting wrong data,11 Hours, -6 hours for downtime "Second and Third Rows". How can I filter JOIN operation in order to get rid of this ? I sorted the two tables in a descending orders, so, the right values come first, but still I’m getting wrong JOINED rows.

Your help is highly appreciated. A.A

Answer 1

Try this. but the results can really differ if you have only one missing record in either of the tables.

Query

SELECT A.MachineID
      ,A.TripTime
      ,B.StartTime
      ,DATEDIFF(HOUR, A.TripTime,B.StartTime) AS DownTime
FROM 
  (
  SELECT * 
    ,ROW_NUMBER() OVER (PARTITION BY MachineID ORDER BY TripTime ASC) AS RN
  FROM dbo.MachineTrip
  ) A
      INNER JOIN
  (
   SELECT * 
       ,ROW_NUMBER() OVER (PARTITION BY MachineID ORDER BY StartTime ASC) AS RN
   FROM dbo.MachineStart
   )B
ON A.MachineID = B.MachineID AND A.RN = B.RN

Result Set

╔═══════════╦═════════════════════════╦═════════════════════════╦══════════╗
║ MachineID ║        TripTime         ║        StartTime        ║ DownTime ║
╠═══════════╬═════════════════════════╬═════════════════════════╬══════════╣
║ XA-065    ║ 2014-03-20 09:40:00.097 ║ 2014-03-20 12:40:00.097 ║        3 ║
║ XA-065    ║ 2014-03-24 18:33:00.040 ║ 2014-03-24 20:33:00.040 ║        2 ║
║ XA-070    ║ 2014-03-21 15:17:00.023 ║ 2014-03-21 18:17:00.023 ║        3 ║
║ XA-765    ║ 2014-03-23 22:16:00.003 ║ 2014-03-23 23:16:00.003 ║        1 ║
╚═══════════╩═════════════════════════╩═════════════════════════╩══════════╝

Working SQL FIDDLE

Suggestion

Your schema needs some serious attention. You should really have a foreign key constraint between these two tables which binds a record in one table to a record in other table.

OR

You can have all the records in one table with a column which binds two related rows together and a column (Maybe a bit column) which indicates whether a record is starttime or triptime.

Your schema as it is very much error prone.

Answer 2

You can also do this with join and aggregation. This is a bit more robust if you have missing records or multiple start records in a row.

select mt.MachineId, mt.TripTime, min(ms.StartTime) as StartTime,
       datediff(hour, mt.TripTime, min(ms.StartTime)) as HoursDifference
from MachineTrip mt join
     MachineStart ms
     on mt.MachineId = ms.MachineId and
        mt.TripTime < ms.StartTime
group by mt.MachineId, mt.TripTime;

Note I used the same naming conventions as Ali so you can try it on that SQL Fiddle.