tags:
- time series
- time series classification
- monster
- EEG
pretty_name: STEW
license: cc-by-4.0
size_categories:
- 10K<n<100K
Part of MONSTER: https://arxiv.org/abs/2502.15122.
| STEW | |
|---|---|
| Category | EEG |
| Num. Examples | 28,512 |
| Num. Channels | 14 |
| Length | 256 |
| Sampling Freq. | 128 Hz |
| Num. Classes | 2 |
| License | CC BY 4.0 |
| Citations | [1] [2] |
STEW comprises raw EEG recordings from 48 participants involved in a multitasking workload experiment [1]. Additionally, the subjects' baseline brain activity at rest was recorded before the test. The data was captured using the Emotiv Epoc device with a sampling frequency of 128Hz and 14 channels, resulting in 2.5 minutes of EEG recording for each case. Participants were instructed to assess their perceived mental workload after each stage using a rating scale ranging from 1 to 9, and these ratings are available in a separate file. The dataset has been divided into cross-validation folds based on individual participants. Additionally, binary class labels have been assigned to the data, categorizing workload ratings above 4 as "high" and those below or equal to 4 as "low". We utilize these labels for our specific problem. STEW can be accessed upon request through the IEEE DataPort [2]. The processed dataset consists of 28,512 multivariate time series each of length 256 (i.e., representing 2 seconds of data at 128 Hz).
[1] Wei Lun Lim, Olga Sourina, and Lipo Wang. (2018). STEW: Simultaneous task EEG workload data set. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 26(11):2106–2114.
[2] Wei Lun Lim, Olga Sourina, and Lipo Wang. (2020). STEW: Simultaneous task EEG workload data set. https://ieee-dataport.org/open-access/stew-simultaneous-task-eeg-workload-dataset. CC BY 4.0.