- Training and testing datasets
The following two datasets will be used for participants to train their model and for contest evaluation:
- Manhattan dataset which is a subset of the New York TLC Trip Record YELLOW Data with only the records such that both pick-up and drop-off are within the Manhattan area. Notice that only the data before July of 2016 has pick-up and drop-off coordinates and therefore is usable for the contest. The Manhattan area is defined by the boundary specified in this KML file. The agent cardinality can be set in a range of 5000-10000. The test dataset will be a few random days of 2016 for the same area.
- Lyft dataset (to be available).