Romain Tavenard
Romain Tavenard
It would be nice to make kNN searches faster when Sakoe-Chiba constrained DTW is concerned using LB_Keogh based pre-filtering. This should be implemented in the `kneighbors` method of class `KNeighborsTimeSeriesMixin`...
**Is your feature request related to a problem? Please describe.** In #262 a new matrix profile implementation is investigated. It has been decided that the multivariate case should not be...
@choltz95 you opened a PR about Gaussian Processes a while ago. I did not have time to deal with it and the PR closed automatically when I removed the dev...
As discussed in #307, several similarity measures for time series (based on softDTW, namely "sharp softDTW" and "softDTW divergence") are introduced in [1] and it would make sense to make...
I received the following question by email* > Dear Romain > > thanks for this toolkit. Can TSlearn handle missing data - quite a big problem in time series analysis...
**Describe the bug** When fed with time series of different dimensionalities, DTW (and other methods that are not designed to handle heterogeneous time series (ie time series whose features do...
At the moment, we use absolute imports in all our codebase and each `tslearn` subpackage is just a single file. I think we should fix these two shortcomings at some...
It would make sense to have metric learning algos dedicated to time series in `tslearn`. A good start could be [Garreau et al, 2014](https://arxiv.org/abs/1409.3136), but maybe other methods could make...
It would be nice to have a method such as BOSS (cf [here](https://www2.informatik.hu-berlin.de/~schaefpa/boss/)) implemented in `tslearn`. Local feature extraction step should be implemented as a `TransformerMixin`.
Keras3 is out and constitutes a significant API change. Typically, our `shapelets` backend is not compatible with keras3 as such, so we need to either: * keep things as such,...