aeon
aeon copied to clipboard
A toolkit for machine learning from time series
What does this implement/fix? Explain your changes. Checking for valid algorithm types #1952 ##### For all contributions - [x] I've added myself to the [list of contributors](https://github.com/aeon-toolkit/aeon/blob/main/.all-contributorsrc). Alternatively, you can...
#### Reference Issues/PRs Fixes #2108 #### What does this implement/fix? Explain your changes. Implements `PyODAdapter` for `IsolationForest` in `anomaly_detection` along with some test cases. #### Does your contribution introduce a...
#### Reference Issues/PRs Issue: #1637 #### What does this implement/fix? Explain your changes. Implemented LSTM-AD algorithm for time series anomaly detection. #### Does your contribution introduce a new dependency? If...
### Describe the bug more legacy nonsense. The functions geometric_mean_absolute_error and geometric_mean_squared_error contain untested code ``` output_errors = _weighted_geometric_mean( np.abs(errors), sample_weight=horizon_weight, axis=0, ) ``` but the actual signature of _weighted_geometric_mean...
### Describe the bug some legacy code in weighted_metrics allows None arguments for the weights, then tests weights.ndim, so crashes if you dont pass weights. I would KWF but its...
### Describe the feature or idea you want to propose Density Peaks (DP) has been used in the TSCL literature. In a recent benchmark paper it was used: https://www.sciencedirect.com/science/article/pii/S2666827020300013. It...
### Describe the feature or idea you want to propose The R-Clustering clusterer is a recently proposed TSCL algorithm using random convolutional kernels (Rocket) with a dimensionality reduction algorithm with...
### Describe the feature or idea you want to propose USSL is a feature based TSCL algorithm that achieves state-of-the-art performance. The paper that proposes it is linked here: https://ieeexplore.ieee.org/document/8386690...
### Describe the feature or idea you want to propose SOM has been used with DTW in the literature and is easily extendable to all the elastic distances we have...
### Describe the feature or idea you want to propose A popular clusterer in the TSCL literature is the k-Spectral Centroid (k-SC). A link to the paper can be found...