1. Although the SampleSubmission.csv period is from 01-01-19 to 01-01-20, the actual test period that is being scored if from 07-01-19 to 12-31-19 inclusive?
2. In the discussion link below, there seems to be an issue where the external metric is not matching with the last specified location before a crash. It is simply matching with the rows in the given order. Does that mean that it will simply match with the first X rows provided in the submission file and not the actual predictions that are made?
- Link: https://zindi.africa/competitions/uber-nairobi-ambulance-perambulation-challenge/discussions/4797
Please clarify or provide additional information. Thank you.
For each crash in the crash period, the scoring looks for the most recent location specified and calculates distance based on that. The rows need to be in date order since the metric does something like `row = sub.loc[sub.date < date].tail(1)`.
If the sample submission is only showing a day then it's just illustrating the format, yours should follow that for the whole test period.