Does thingworx have MLOps capabilities for champion-challenger experiments?

[LR_9796586]

I have to develop a solutions that involve supervision of the model performance over time, and compare the production model against other models with different algorithms or trained with different data.

The flow could be analogous to the below image, were the Production Model is the main one, but if any of the other models gets better results, e.g. Challenger 3 then it takes the position as the new champion. 

[cmorfin · Accepted solution]

Hi @LR_9796586 

There is no such functionality in ThingWorx Analytics.

If the models are stored in ThingWorx Analytics, there is possibly ways to achieve this but that will require you to write services to compare the performance of each models and then use the best one for scoring new data.

You can also post a Product Idea to have this type of functionality added in the product in future release.

Christophe

[slangley]

Hi @LR_9796586.

If you feel your question has been answered, please mark the previous response as the Accepted Solution for the benefit of others with the same question.

Regards.

--Sharon