tags:
- time series
- time series classification
- monster
- other
- sensor
license: other
pretty_name: FordChallenge
size_categories:
- 10K<n<100K
Part of MONSTER: https://arxiv.org/abs/2502.15122.
| FordChallenge | |
|---|---|
| Category | Sensor |
| Num. Examples | 36,257 |
| Num. Channels | 30 |
| Length | 40 |
| Sampling Freq. | 10 Hz |
| Num. Classes | 2 |
| License | Other |
| Citations | [1] |
FordChallenge is obtained from Kaggle and consists of data from 600 real-time driving sessions, each lasting approximately 2 minutes and sampled at 100ms intervals [1] (i.e., a sampling rate of 10 Hz). The processed dataset consists of 36,257 multivariate time series each of length 40 (i.e., representing 4 seconds of data per time series at 10 Hz). These sessions include trials from 100 drivers of varying ages and genders. The dataset contains 8 physiological, 11 environmental, and 11 vehicular measurements, with specific details such as names and units undisclosed by the challenge organizers. Each data point is labeled with a binary outcome: 0 for "distracted" and 1 for "alert". The objective of the challenge is to design a classifier capable of accurately predicting driver alertness using the provided physiological, environmental, and vehicular data.
[1] Mahmoud Abou-Nasr. (2011). Stay Alert! The Ford Challenge. https://kaggle.com/competitions/stayalert. Kaggle.