Using river with offline ML models

[RafiullahOmar]

Can we use River with offline ML models to detect concept drift? Additionally, where can I find code examples for concept drift detection using river in offline ML models with a real and Synthetic datasets?

[MaxHalford]

Hey there.

No we don't support offline models. But then again its not very clear what you're trying to do. You haven't provided a lot of details.

And this is River, so we can't help you with scikit-multiflow. We do have real and synthetic datasets though.

[RafiullahOmar]

Dear @MaxHalford, Apologies for the typing error in my question, where I mistakenly mentioned scikit-multiflow. I am seeking a complete coding example for concept drift detection using the River library.
To provide more context, my goal is to train an offline model on the first half of a dataset. Then, I want to generate a stream of data from the other half of the dataset and apply various concept drift detection methods available in the River library to detect concept drift. And where I can find the compete documentations of river, as there are less details in river website.
I appreciate your assistance and clarification regarding the use of River for this specific scenario.

[smastelini]

@RafiullahOmar, you should check the example sections of the detectors in the documentation (for instance, here) to learn about the specific usage of a detector.

Moreover, you can check the drift detection guide for more details on this type of task. Aside from that, I suggest you check the literature on concept drift detection to understand more about this phenomenon.