topic Re: Miner - HP Forest predecessor node in SAS Data Science

Miner - HP Forest predecessor node

DarioM — Wed, 13 Jul 2022 19:36:24 GMT

Wondering why this is my error and how to fix it. “Must use at least one input or rejected variable” for node HP Forest.

Previous node is score, and before that I have 20 LARS’s connected by 20 Sample nodes for marketing analysis.

Re: Miner - HP Forest predecessor node

sbxkoenk — Wed, 13 Jul 2022 20:10:04 GMT

Hello,

Which Enterprise Miner version?

EM 13.2 or EM 14.1 or EM 15.2 or ??

Thanks,

Koen

Re: Miner - HP Forest predecessor node

DarioM — Wed, 13 Jul 2022 21:19:04 GMT

14.1.

Re: Miner - HP Forest predecessor node

DarioM — Thu, 14 Jul 2022 13:23:14 GMT

here is my diagram. response rate is < 1%.. but i get the error above as previously mentioned.

Re: Miner - HP Forest predecessor node

sbxkoenk — Thu, 14 Jul 2022 21:45:28 GMT

Hello,

No idea why you get this error.

Are you sure the input data source to the HP Forest node has the role of TRAIN or RAW?

What happens when you start a new process flow (pipeline) with HP Forest node as second or third node (after data partitioning)?

Maybe best to contact SAS Technical Support in your country if the problem persists.
Technical Support can set up a call with screen sharing and then the root cause will be determined quickly.

Thanks,

Koen

Re: Miner - HP Forest predecessor node

GuyTreepwood — Thu, 14 Jul 2022 21:53:58 GMT

Did you check to see if any variables are being fed into the HP Forest 2 node from the HP Data Partition node? I believe you would need to check the 'Exported Data' property option. If do not see any variables flowing through that node, then you might need to back track and check the nodes that are run before the HP Data Partition.

Re: Miner - HP Forest predecessor node

DarioM — Thu, 14 Jul 2022 22:25:26 GMT

Not sure how to check if it has TRAIN role?

Maybe I should be partitioning before I run the many 20 samples?

I added an append node before partition and it ran but my results are weird. I’m from Canada, not sure where to get on a screen share with admin

Re: Miner - HP Forest predecessor node

DarioM — Thu, 14 Jul 2022 22:28:19 GMT

Maybe I should have the partition before the oversample? Can this be contributing to it? Not sure how to have my 20 separate samples go into 1 dataset that can then be used to run a forest model. Maybe partition should come after or before? Not sure. Going by what I saw in SAS videos, but the node that they showed does not exist in 14.2.

Re: Miner - HP Forest predecessor node

sbxkoenk — Fri, 15 Jul 2022 09:42:01 GMT

SAS Technical Support ( free of charge! )
https://support.sas.com/en/technical-support.html

Contact SAS Technical Support
https://www.sas.com/en_us/contact/technical-support.html

For more tips about contacting SAS Technical Support, read SAS Note 57691.

SAS Tutorial | 4 Tips for Working with SAS Technical Support
Jan 25, 2021
https://www.youtube.com/watch?v=1cfw9Wi4R6I

Koen

Re: Miner - HP Forest predecessor node

DarioM — Fri, 15 Jul 2022 12:17:11 GMT

I do not see any variables flowing through. Not sure how to get the 20 samples into one to run the forest model. Should I be partitioning the data before oversampling?

Re: Miner - HP Forest predecessor node

DarioM — Fri, 15 Jul 2022 12:40:16 GMT

Yes my forest node is the 4th node and it is after partition, but i am not sure if this logic is correct, although i got it from this video https://www.youtube.com/watch?v=jeZX31XsmFs not sure if partitioning after is what is the correct thing to do. i contacted them and am hoping for a meeting shortly. im pretty sure this append node is not doing the job correctly.