Need Aggregation logic in transformer

[dstest]

I have areuqirement like below.

I need to find the max value by grouping on two columns.If even if i have a negative value but i need to take absolute value but in the output i need to populate originaal value.



Input data:

100,200,-5
100,200,1
100,200,-4

Output :
100,200,-5

Thanks

[stefanfrost1]

You need to ensure sorted data by your defined keys and ensure that your partitioning supports that sorted order. When all that is complete then you can create one or as many stage varibles as needed to kep track of your keys and stage varibles for your aggregates or max-values. Then just let datastage process your data only letting the last record of each group go through the transformer carrying the aggregate values. Set changedrec key at your sort to simplify this process and make sure it defines the last record in your group...

Simple example

IsSame = If LastKey = InStream.Key Then Yes Else No
LastKey = InStream.Key
Counter = If IsSame = Yes Then Counter + 1 Else 0
ValueCount = If IsSame = Yes Then ValueCount + InStream.Value Else 0
MaxValue =
If IsSame = Yes Then
If MaxValue < InStream.MaxValue Then InStream.MaxValue
Else MaxValue
Else -999

In Your Output Stream, Put Filter Option
If IsSame = Yes And LastRec = 1 Then 1 Else 0

[stefanfrost1]

You need to ensure sorted data by your defined keys and ensure that your partitioning supports that sorted order. When all that is complete then you can create one or as many stage varibles as needed to kep track of your keys and stage varibles for your aggregates or max-values. Then just let datastage process your data only letting the last record of each group go through the transformer carrying the aggregate values. Set changedrec key at your sort to simplify this process and make sure it defines the last record in your group...

Simple example

IsSame = If LastKey = InStream.Key Then Yes Else No
LastKey = InStream.Key
Counter = If IsSame = Yes Then Counter + 1 Else 0
ValueCount = If IsSame = Yes Then ValueCount + InStream.Value Else 0
MaxValue =
If IsSame = Yes Then
If MaxValue < InStream.MaxValue Then InStream.MaxValue
Else MaxValue
Else -999

In Your Output Stream, Put Filter Option
If IsSame = Yes And LastRec = 1 Then 1 Else 0

[stefanfrost1]

You need to ensure sorted data by your defined keys and ensure that your partitioning supports that sorted order. When all that is complete then you can create one or as many stage varibles as needed to kep track of your keys and stage varibles for your aggregates or max-values. Then just let datastage process your data only letting the last record of each group go through the transformer carrying the aggregate values. Set changedrec key at your sort to simplify this process and make sure it defines the last record in your group...

Simple example

IsSame = If LastKey = InStream.Key Then Yes Else No
LastKey = InStream.Key
Counter = If IsSame = Yes Then Counter + 1 Else 0
ValueCount = If IsSame = Yes Then ValueCount + InStream.Value Else 0
MaxValue =
If IsSame = Yes Then
If MaxValue < InStream.MaxValue Then InStream.MaxValue
Else MaxValue
Else -999

In Your Output Stream, Put Filter Option
If IsSame = Yes And LastRec = 1 Then 1 Else 0

[stefanfrost1]

Sorry about the multiple posting

[chulett]

Next time, know that you can always delete your own post as long as there are no replies to it. So when this happens to me, I start from the bottom and delete the extra ones. Too late now, though.

[DSguru2B]

A suggestion: Instead of writing all that code in the transformer, you can create a temp table, load your data with an extra column that has absolute values, do a group by and get your result.