https://github.com/borgwardtlab/uea_ucr_datasets
A small package for loading and handling UEA UCR time series classification datasets.
https://github.com/borgwardtlab/uea_ucr_datasets
Last synced: 11 months ago
JSON representation
A small package for loading and handling UEA UCR time series classification datasets.
- Host: GitHub
- URL: https://github.com/borgwardtlab/uea_ucr_datasets
- Owner: BorgwardtLab
- Created: 2019-11-05T10:56:22.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2020-08-07T14:51:51.000Z (almost 6 years ago)
- Last Synced: 2024-11-09T12:18:43.591Z (over 1 year ago)
- Language: Python
- Size: 15.6 KB
- Stars: 5
- Watchers: 4
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# uea_ucr_datasets
This package contains convenience functions and classes to access the UEA UCR
time series classification archive.
Currently it contains the following functionalities:
- `Dataset` class: Loads UEA UCR dataset stored in the `sktime` format
from `~/.data/UEA_UCR/` or path provided via the `UEA_UCR_DATA_DIR`
environment variable. This class is compatible with the pytorch `DataLoader`
class.
- `list_datasets`: List datasets available in the `~/.data/UEA_UCR/` folder or
path provided via the `UEA_UCR_DATA_DIR`
## Example usage
Download the `sktime` version of the UEA and UCR datasets. And unpack them.
Move the folders of the individual datasets to the path `~/.data/UEA_UCR`.
```python
>>> import uea_ucr_datasets
>>> uea_ucr_datasets.list_datasets()
['LSST',..]
>>> d = uea_ucr_datasets.Dataset('UWaveGestureLibrary', train=True)
>>> first_instance = d[0]
>>> instance_x, instance_y = first_instance
```
## Alternative data paths
You can also store the data at another location, then it is required to set the
environment variable `UEA_UCR_DATA_DIR` appropriately such that the package can
find the datasets.