[TRANSLATION ERROR] There are no messages for this key

Next Place Prediction API Tutorials

The Sherpa.ai Next Place Prediction API is a REST API that provides developers with a framework and toolbox to formulate any request for Sherpa.ai, by using standard syntax technologies like HTTP Rest Services with a predefined set of URLs. You can use it to create a prebuilt recommender just by using these endpoints.

Sherpa.ai has created five datasets, so that the spatial context, profiling, and prebuilt recommendation functionalities of the API can be tested.

The datasets simulate a user’s route, points of interest, and time transitions (the amount of time a user spends at a point of interest, arrival time, and departure time). That way, the points of interest that the user frequents can be detected, based on their location, the date, and the time.

The datasets simulate a user’s routine over the course of one month (February 2020), in the the following cities:

  • San Francisco
  • New York
  • Madrid
  • London
  • Singapore

In each dataset, the user has points of interest (POI) that can be categorized, as follows:

  • Home
  • Office
  • Leisure

The time spent by the user at each point of interest is simulated, using:

  • Arrival time
  • Departure time

Additionally, we have established three sections within the general routine:

  • Monday to Thursday
  • Friday
  • Saturday and Sunday

These sections and routines have been created using the following basis and variability:

  • The Home and Office POIs are initially predefined.
  • The arrival time at the Work POI is a random time each day, between 8:00 and 9:00 AM, from Monday to Friday.
  • The departure time from the Work POI is a random time each day, between 5:00 and 6:00 PM, from Monday to Thursday; and between 3:00 and 4:00 PM, on Friday.
  • In addition to Home and Work POIs, the user is encouraged to occasionally visit a Leisure POI from Monday to Thursday. However, on Friday, the user is allowed to visit as many Leisure POIs as they would like, though they are encouraged not to exceed a random Home POI arrival time of 11:00 PM.
  • The time spent at Leisure POIs is an arbitrary time, between 30 and 60 minutes from Monday to Friday; and between one and four hours, on weekends.
  • Arrival and departure times to and from the Home POI are governed by the number of Leisure POIs visited and/or routing travel time for Monday to Thursday. On Fridays, users are encouraged, but not restricted, to arrive at the Home POI between 9:00 and 11:00 PM. On weekends, arrival time ranges from 7:00 to 11:00 PM, and departure time ranges from 9:00 AM to 3:00 PM.
  • The routes between POIs have been simulated, based on a normal car trip between points. The speed of the car depends upon the road and ranges between 10 km/h and 60 km/h. However, the data accounts for the decrease in speed, in relation to traffic lights, intersections, and U-turns.
  • Sampling time for points generated at POIs starts from as low as 30 seconds, while for routes, it starts from as low as one second.

See the Testing Datasets section, to analyze the technical details of the data.