Implementation of McDiarmid Drift Detection Methods?

[aditya0by0]

Hi everyone,

I recently came across the paper:

Pesaranghader, Ali, et al. "McDiarmid Drift Detection Methods for Evolving Data Streams", International Joint Conference on Neural Networks, 2018.
Pre-print: https://arxiv.org/abs/1710.02030
DOI: https://doi.org/10.1109/IJCNN.2018.8489260

This paper introduces a McDiarmid-based drift detection method for evolving data streams. I was wondering:

1. Is there an existing implementation of this method in River?
2. If not, would it be useful to implement it?
3. Are there any alternative drift detection methods in River that are similar to McDiarmid?

Any insights or suggestions would be greatly appreciated!

Thanks!

[smastelini]

Hi @aditya0by0, thanks for your interest! I was not familiar with this specific detector.

Regarding your questions:

1. No
2. Given the amount of citations and (a brief) look in the experimental evaluation in the paper, I would say, yes!
3. Many of the algorithms compared against the McDiarmid drift detector are available in River :D

[aditya0by0]

Thanks for reply, I would love to implement it. Give me sometime and I can create a PR for it.

[smastelini]

Cool! We will wait for it!

[aditya0by0]

@smastelini, Can you please check my PR (#1703) for this discussion and help me out . I have some comment for it.