How to set a seed for ThingWorx analytics models

Forum|Forum|4 years ago
May 6, 2021
2 replies
1080 views

I'm training Models through the UI and the API and I'm getting slightly different results each time, even when the parameters are the same the results are not replicable.

In particular I'm training a logistic regression, an algorithm that should be deterministic, I'm also setting the training data to 100% and 0% for validation so there should no be an effected of the sampling there.

Is there an option to set a seed or any other mechanist to get the exact same result each time?

Analytics

Best answer by jgreiner

Hello,

I heard back for the AI team and there is no way for a user to set a seed for a model as the seed is not shared with end users. The reason why you are getting different results for logistic regression is that our algorithm is perceptron based and not deterministic. If you are looking for deterministic model, then a linear regression or decision should be used.

Let me know if you have any further questions or concerns.

Warm Regards,

John

J

jgreiner

13-Aquamarine

Hello,

I will reach out to our AI team to get a definitive answer to your questions. Let me know if you have any questions or concerns in the meantime.

Warm Regards,

John

J

jgreiner

Answer

13-Aquamarine

Hello,

I heard back for the AI team and there is no way for a user to set a seed for a model as the seed is not shared with end users. The reason why you are getting different results for logistic regression is that our algorithm is perceptron based and not deterministic. If you are looking for deterministic model, then a linear regression or decision should be used.

Let me know if you have any further questions or concerns.

Warm Regards,

John

Sign up

Please use your PTC eSupport account.

Welcome to the PTC Community

Please use your PTC eSupport account.

Scanning file for viruses.

This file cannot be downloaded