Skip to main content
T
trungkien21041-Visitor
1-Visitor
September 14, 2019
Solved

Learner in Thingworx analytics

  • September 14, 2019
  • 4 replies
  • 1767 views

Hello,

I'm a newbie in thingworx analytics but I come from data analytics. Actually, the learner is not new for me but when I configure the predictive model, I saw that I can add more leaners. So, that means the analytics module shall try to use these methods to learning the data and choose the best fit (highest ROC, RMSE, etc.) method ??? If not, what does the module do exactly ????

 

Best answer by nsampat

@trungkien2104 ,

 

Your understanding of the process is correct.

 

If you have any product suggestions, you can make a post on the Ideas Portal, where it can be voted upon by other users if they like the idea.

 

These Ideas are then reviewed by the Product Managers who may use them in future releases.

 

You can find the Ideas Page here: ThingWorx Ideas

 

Also, if possible, can you mark my post above as Solution Accepted if it answers your question?

 

If not, do you require any other clarifications?

 

Regards,

 

Neel

4 replies

N
nsampat17-Peridot
17-Peridot
September 16, 2019

@trungkien2104 ,

 

Thank you for posting your question to the PTC Community.

 

You have the option of using the Learners individually or in an ensemble.

 

You are essentially correct, as it pertains to using an ensemble setup.

 

More information can be found on the ThingWorx Analytics HelpCenter - Learners and Ensemble Techniques page.

 

Please let me know if you have any additional questions.

 

Regards,

 

Neel

T
trungkien21041-VisitorAuthor
1-Visitor
September 17, 2019

Dear Neel, @nsampat 

Thank you for your response.

That mean although we select an ensemble of learners, only a best learner is select by ordering the score. And by choosing the option (Average, Best, Elite Average, Majority Vote and Soloist), we can see the average score. Then, I think PTC can add the result of selected learner (the learner is chosen with the highest value) such as the parameter of model (for example the coefficient  in linear regression). It can help developer to tuning or validate the model with different periods of time.

N
nsampat17-PeridotAnswer
17-Peridot
September 17, 2019

@trungkien2104 ,

 

Your understanding of the process is correct.

 

If you have any product suggestions, you can make a post on the Ideas Portal, where it can be voted upon by other users if they like the idea.

 

These Ideas are then reviewed by the Product Managers who may use them in future releases.

 

You can find the Ideas Page here: ThingWorx Ideas

 

Also, if possible, can you mark my post above as Solution Accepted if it answers your question?

 

If not, do you require any other clarifications?

 

Regards,

 

Neel

    Terms & ConditionsAccessibility statement