Hi,
I'm trying to automate the feature selection process in our machine learning process, dynamically filtering out all variables that have only 1 value and the string ones that have over 32.
I managed to do most of the job using a few basic data profile tools, followed by some filters to exclude unwanted columns and a cross-tab to get it back to columnar format.
My problem now is that this way I loose the original field type and I need to find a way to reassign it before I do an union to actually select the final columns. Is there any way to do it in Alteryx?
Workflow attached.
Thanks in advance!
Solved! Go to Solution.
Hi @marco_zara
There is a macro within the CReW macros that does exactly what you are after.
They can be found here: http://www.chaosreignswithin.com/p/macros.html
However looking into CReW_DynamicMetadata I noticed there was a bug/
So please find attached one that should work. Along with it's sub-macro (CReW_DynamicMetadata.yxmcbatch) and your workflow (all be it I had to change your input data tool)
@MarqueeCrew - The bug is with formula 17 in the macro. It was:
iif(isnull([Scale]), tostring([Scale]), tostring([Size]) + "." + tostring([Scale]))
but should be:
iif(isnull([Scale]), tostring([Size]), tostring([Size]) + "." + tostring([Scale]))
Hi marco_zara,
One option is that you can save the field configuration of your input as a .yxft file and then load that in a Select tool to replace the new field types to the original field types.
Save the field configuration in the select tool like below.
This will prompt you to save a .yxft file. Then, simply select Load Field Names & Types to update the Select tool.
The only problem with this is that it is not a dynamic process so let me know if you except the field names and types to change frequently.
Josh
Thank you very much, this should solve the issue pretty effectively!