Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

Is it possible to filter outputs in Hyperstudy?

Hi!

I am using Hyperstudy 2025 and I am doing an IPT optimization in EDEM. Sometimes the simulation "fails" and the output then becomes 0(see figure). I have been experimenting with machine learning to save time for all the simulations, but when some simulations "fail" and the output is read as 0 it affects the machine learning and it creates a wrong model. So I was thinking if it was possible to filter out these outputs automatically that become 0 in Hyperstudy so I can use a "sampling fit model"?

BR

Find more posts tagged with

HyperStudy

Accepted answers

All comments

GTT Adam

Hi @Karl_H,

In HyperStudy, you can choose to exclude failed runs:

Hope this helps!

Adam Reid

Karl_H

Hi! The problem is that it isn't failed run, instead the output reads 0 so Hyperstudy can't see that it failed and instead thinks the output was 0. So SFA reads 0 but it should just ignore those values instead.

GTT Adam

Hi @Karl_H,

You can set a constraint on an Output Response in the Setup such that the result must be greater than 0:

Hope this helps!

Adam Reid

Garima_Singh

Hi @Karl_H,

Recommendations for the HyperStudy query are as follows:

Since, the requirement is to create a mathematical/predictive model, it is NOT recommended to use the data points created from an optimization method. It is recommended to use MELS DOE method for the same.
For creating the mathematical/predictive model, either one can perform a MELS DOE & then use 'Fit Existing Data' approach or else, one can use the 'Sampling FIT' approach which performs MELS DOE & FIT model creation in a single-step. To use 'Sampling FIT', one needs to mention the R2 value (accuracy) required for the predictive model as one of the inputs.

To understand the 'Fit Existing Data' & 'Sampling FIT' approach, kindly refer the HyperStudy e-learning on the same (as shown below):

HyperStudy e-learning https://learn.altair.com/courses/getting-started-with-hyperstudy-elearning

Now, regarding filtering the iterations having SFA as 0:

Unless the iteration fails, HyperStudy cannot ignore the iterations with SFA as 0 by default.
Hence, once a DOE approach is performed, before creating the 'FIT' model, one can use the 'Edit' option to deleted the iterations with SFA as 0 using 'Remove Run' option (as shown below).

Note:

It is recommended to only use 'Fit Existing Data' for your requirement as the iterations do not fail when SFA is 0.
In an Optimization approach, by setting a constraint of SFA > 0, the optimizer will ignore the iterations wherein the combinations of DV's tend to make SFA as 0. But, it is not recommended to use data from an optimization for creating a predictive model (as explained in the above section).

Hope that the insights provided serve useful.

Many thanks.

Kind Regards

Garima Singh

Engineering Data Science