RacketSports

Task: Classification Classes: 4 Channels: 6 Length: 30

Overview

Multivariate time series classification dataset from the UEA archive (Bagnall et al., 2018). University students played badminton or squash while wearing a smart watch, which recorded the x, y and z axes of its gyroscope and its accelerometer (6 channels) at 10 Hz for 3 seconds. The task is to tell the sport and the stroke: a clear or a smash in badminton, a forehand or a backhand boast in squash. The data was collected by Phillip Perks as part of an undergraduate project at the University of East Anglia in 2017/18.

../../_images/racket_sports_teaser.png

The archive does not document which channels belong to the gyroscope and which to the accelerometer, so the channels are kept in the archive’s order. The loader passes the archive’s data through unchanged.

  • Train: 151 recordings.

  • Test: 152 recordings.

Class names are the archive’s labels in lower case: badminton_clear, badminton_smash, squash_backhandboast, squash_forehandboast.

Data Structure

Key

Type

Description

series

list of list of float32

Smart watch recording, shape (30, 6): time steps by channels

label

int

Stroke class (0-3)

Usage Example

import numpy as np
from stable_datasets.timeseries import RacketSports

train = RacketSports(split="train")
sample = train[0]
x = np.asarray(sample["series"])  # (30, 6)
print(x.shape, train.features["label"].names[sample["label"]])

References

Citation

@article{bagnall2018uea,
  title={The UEA multivariate time series classification archive, 2018},
  author={Bagnall, Anthony and Dau, Hoang Anh and Lines, Jason and Flynn, Michael and Large, James and Bostrom, Aaron and Southam, Paul and Keogh, Eamonn},
  journal={arXiv preprint arXiv:1811.00075},
  year={2018}
}