The viewer is disabled because this dataset repo requires arbitrary Python code execution. Please consider
removing the
loading script
and relying on
automated data support
(you can use
convert_to_parquet
from the datasets library). If this is not possible, please
open a discussion
for direct help.
Part of MONSTER: https://arxiv.org/abs/2502.15122.
| Pedestrian | |
|---|---|
| Category | Count |
| Num. Examples | 189,621 |
| Num. Channels | 1 |
| Length | 24 |
| Sampling Freq. | hourly |
| Num. Classes | 82 |
| License | CC BY 4.0 |
| Citations | [1] |
Pedestrian represents hourly pedestrian counts at 82 locations in Melbourne, Australia between 2009 and 2022 [1]. The processed dataset consists of 189,621 (univariate) time series, each of length 24 (i.e., representing 24 hours of data per time series). The data comes from automatic pedestrian counting sensors at different locations. The task is to identify location based on the time series of counts. The dataset has been split into stratified random cross-validation folds.
[1] City of Melbourne. (2022). Pedestrian counting system. https://data.melbourne.vic.gov.au/explore/dataset/pedestrian-counting-system-monthly-counts-per-hour/information/. CC BY 4.0.
- Downloads last month
- 172