Given a set of counts at locations per n time periods, the algorithm will forecast the number at time n+1
Applicable Scenarios and Problems
Forecasting counts of an event (e.g. numbers of disease cases, numbers of tests used) where locations are known and data are available for multiple time periods.
Requires nested json of 2 tables.
The first table should be named 'count_data' and should contain 3 fields:
- 'site' a numeric or character ID for each site
- 'count' the number of the outcome/events
- 'time' the time period in whole integers. These don’t have to be equally spaced.
The second table should be named 'coordinates' and should contain 3 fields:
- 'site' a numeric or character ID for each site. Must contain all the sites present in the 'count_data' table
- 'lng' the longitude of each site
- 'lat' the latitude of each site
A json containing 2 fields
- 'site' the site ID
- 'forecast' the predicted number of events in time n+1