Principal Components Analysis / Batch Macro

[robsuddaby]

I've created a control parameter based batch macro to run a PCA analysis based upon 2 fields which change depending on certain parameters. However, if I feed in 10 fields it returns me 100 records (10x10) and takes ~30s to run. I can summarise the data afterwards and get the results I want no problem but running the actual PCA tool against 1 set of parameters will return my 10 records in 2s with the same results... however, I actually need to run ~35k records in my live scenario so the macro is not feasible.

I've tried toying with Control GroupBy and Control Parameter Batch Macros but neither seem to work the way I want; which is essentially to group my PCA analysis (containing say 10 records) by my 2 fields (which can be combined into 1 for ease if needs be). Any thoughts? 

[dsmdavid · Accepted solution]

Hi Rob, 
I'm a bit confused with what are your expectations here for the batch macro, so maybe my answer doesn't make a lot of sense.

As I see the workflow right now, you're feeding...
1. 10 rows of data.

2. 10 rows to field1 (all with a value = 9).

3. 10 rows to field2 (all with value = type).

The outcome of that is that you have 10 x 10 iterations (10 field1 x 10 field2), all of them running with the same values, so they're identical.

An alternative could be to split the feed of data & parameters, for instance having a group by immediately before the control parameter:

I've modified the macro to include the value of the fields used to debug in the attached wf.

[robsuddaby]

Thanks very much David, I did not realise you can 'fiddle' with the inputs to a batch macro - very neat and simple solution!