Detection advice

[RobertoDF]

Hi, I've been trying different search methods and models on the following dataset. For example:

model = "rank"  # "l1", "rbf", "linear", "normal", "ar",...
algo = rpt.Binseg(model=model, jump=50).fit(behavior["Running speed"].rolling(1000,center=True).mean().values)
limits = algo.predict(pen=1000)

I couldn't reach a good enough result, can somebody point at what would be a good direction? test_data_pickle

[oboulant]

Hi @RobertoDF ,

Thanks for your interest in ruptures !

Closing this since it is not an issue, rather a discussion. Feel free to open a thread in the discussion section.

[RobertoDF]

Sorry for the misplacement!