Dynamic geofence zones for ride sharing

1. A computer-implemented method, comprising:

creating, by one or more servers, a static geofence fixed to a geographic location;

in response to collecting rider behavior data and safety incident data over a first threshold, converting, by the one or more servers based on the collected data, the static geofence to at least one set of dynamic geofence zones, the at least one set of dynamic geofence zones corresponding to a plurality of spatial-temporal cubes, wherein each cube represents a dynamic geofence zone, a spatial dimension of the cube comprises a geohash code corresponding to an adjustable area within the geographic location, and a temporal dimension of the cube comprises at least one time period, and the at least one set of dynamic geofence zones are based at least on residence addresses of flagged riders;

deploying, by the one or more servers, the at least one set of dynamic geofence zones for responding to a trip request, wherein the responding to the trip request comprises:

respectively based on a rider corresponding to the trip request, selecting a number of one or more of the at least one set of dynamic geofence zones and a size of each of the one or more of the at least one set of dynamic geofence zones;

based on the collected rider behavior data and safety incident data, training, by the one or more servers, a machine learning model to predict one or more areas to have a number of safety incidents in the future over a second threshold;

updating, by the one or more servers, the plurality of cubes corresponding to the at least one set of dynamic geofence zones, wherein the updating comprises:

moving, based on the trained machine learning model and residence address change of one or more of the flagged riders, one dynamic geofence zone from an area to a different area predicted to have a number of safety incidents in the future over the second threshold by changing a geohash location in the cube corresponding to the one dynamic geofence zone, and

adjusting the cube by (i) adjusting, based on a model recall accuracy of the one dynamic geofence zone, a number of characters of the geohash code of the cube corresponding to the one dynamic geofence zone to increase or reduce a size of the one dynamic geofence zone in the spatial dimension and (ii) shifting or changing a length of the at least one time period in the temporal dimension;

receiving, by the one or more servers, a request for a trip from a computer device of the rider;

determining, by the one or more servers, that a pickup location is within the one dynamic geofence zone of the at least one set of dynamic geofence zones and that a time of the trip is within a time period of the one dynamic geofence zone of the at least one set of dynamic geofence zones; and

based on the determination, transmitting, by the one or more servers, a request to the computer device of the rider requiring the rider to make a new request from a different location from the pickup location.

2. The method of claim 1 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones further comprises removing at least one dynamic geofence zone from the at least one set of dynamic geofence zones.

3. The method of claim 1 , wherein adjusting the number of characters of the geohash code of the cube corresponding to the dynamic geofence zone comprises reducing a size of the dynamic geofence zone by increasing the number of characters of the geohash code of the cube corresponding to the dynamic geofence zone.

4. The method of claim 1 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones further comprises changing the at least one time period of the one dynamic geofence zone.

5. The method of claim 1 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones comprises changing the geohash code of a cube corresponding to at least one dynamic geofence zone of the at least one set of dynamic geofence zones.

6. The method of claim 5 , wherein:

the one dynamic geofence zone of the at least one set of dynamic geofence zones is deployed at a residence address associated with a flagged rider; and

the geohash code of the one dynamic geofence zone of the at least one set of dynamic geofence zones is changed to a new residence address associated with the flagged rider.

7. The method of claim 1 , wherein deploying the at least one set of dynamic geofence zones comprising:

obtaining a number of flagged riders in a new area;

determining whether the number of flagged riders is greater than a threshold; and

in response to determining that the number of flagged riders is greater than the threshold, deploying at least one dynamic geofence zone of the at least one set of dynamic geofence zones to the new area.

8. A system, comprising one or more processors and one or more non-transitory computer-readable memories coupled to the one or more processors and configured with instructions executable by the one or more processors to cause the system to perform operations comprising:

creating a static geofence fixed to a geographic location;

in response to collecting rider behavior data and safety incident data over a first threshold, converting, based on the collected data, the static geofence to at least one set of dynamic geofence zones, the at least one set of dynamic geofence zones corresponding to a plurality of spatial-temporal cubes, wherein each cube represents a dynamic geofence zone, a spatial dimension of the cube comprises a geohash code corresponding to an adjustable area within the geographic location, and a temporal dimension of the cube comprises at least one time period, and the at least one set of dynamic geofence zones are based at least on residence addresses of flagged riders;

deploying the at least one set of dynamic geofence zones for responding to a trip request, wherein the responding to the trip request comprises:

determining a remedial action when a pickup location or a destination of the trip request is within the spatial dimension of the cube and a time of the trip request is within the temporal dimension of the cube respectively based on a rider corresponding to the trip request, selecting a number of one or more of the at least one set of dynamic geofence zones and a size of each of the one or more of the at least one set of dynamic geofence zones;

based on the collected rider behavior data and safety incident data, training a machine learning model to predict one or more areas to have a number of safety incidents in the future over a second threshold;

updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones, wherein the updating comprises:

moving, based on the trained machine learning model and residence address change of one or more of the flagged riders, one dynamic geofence zone from an area to a different area predicted to have a number of safety incidents in the future over the second threshold by changing a geohash location in the cube corresponding to the one dynamic geofence zone, and

adjusting the cube by (i) adjusting, based on a model recall accuracy of the one dynamic geofence zone, a number of characters of the geohash code of the cube corresponding to the one dynamic geofence zone to increase or reduce a size of the one dynamic geofence zone in the spatial dimension and (ii) shifting or changing a length of the at least one time period in the temporal dimension;

receiving a request for a trip from a computer device of the rider;

determining that a pickup location is within the one dynamic geofence zone of the at least one set of dynamic geofence zones and that a time of the trip is within a time period of the one dynamic geofence zone of the at least one set of dynamic geofence zones; and

based on the determination, transmitting a request to the computer device of the rider requiring the rider to make a new request from a different location from the pickup location.

9. The system of claim 8 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones further comprises removing at least one dynamic geofence zone from the at least one set of dynamic geofence zones.

10. The system of claim 8 , wherein adjusting the number of characters of the geohash code of the cube corresponding to the dynamic geofence zone comprises reducing a size of the dynamic geofence zone by increasing the number of characters of the geohash code of the cube corresponding to the dynamic geofence zone.

11. The system of claim 8 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones further comprises changing the at least one time period the one dynamic geofence zone.

12. The system of claim 8 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones comprises changing the geohash code of a cube corresponding to at least one dynamic geofence zone of the at least one set of dynamic geofence zones.

13. The system of claim 12 , wherein:

the one dynamic geofence zone of the at least one set of dynamic geofence zones is deployed at a residence address associated with a flagged rider; and

the geohash code of the one dynamic geofence zone of the at least one set of dynamic geofence zones is changed to a new residence address associated with the flagged rider.

14. The system of claim 8 , wherein deploying the at least one set of dynamic geofence zones comprising:

obtaining a number of flagged riders in a new area;

determining whether the number of flagged riders is greater than a threshold; and

in response to determining that the number of flagged riders is greater than the threshold, deploying at least one dynamic geofence zone of the at least one set of dynamic geofence zones to the new area.

15. A non-transitory computer-readable storage medium configured with instructions executable by one or more processors to cause the one or more processors to perform operations comprising:

creating a static geofence fixed to a geographic location;

in response to collecting rider behavior data and safety incident data over a first threshold, converting, based on the collected data, the static geofence to at least one set of dynamic geofence zones, the at least one set of dynamic geofence zones corresponding to a plurality of spatial-temporal cubes, wherein each cube represents a dynamic geofence zone, a spatial dimension of the cube comprises a geohash code corresponding to an adjustable area within the geographic location, and a temporal dimension of the cube comprises at least one time period, and the at least one set of dynamic geofence zones are based at least on residence addresses of flagged riders;

deploying the at least one set of dynamic geofence zones for responding to a trip request, wherein the responding to the trip request comprises:

determining a remedial action when a pickup location or a destination of the trip request is within the spatial dimension of the cube and a time of the trip request is within the temporal dimension of the cube respectively based on a rider corresponding to the trip request, selecting a number of one or more of the at least one set of dynamic geofence zones and a size of each of the one or more of the at least one set of dynamic geofence zones;

based on the collected rider behavior data and safety incident data, training a machine learning model to predict one or more areas to have a number of safety incidents in the future over a second threshold;

updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones, wherein the updating comprises:

moving, based on the trained machine learning model and residence address change of one or more of the flagged riders, one dynamic geofence zone from an area to a different area predicted to have a number of safety incidents in the future over the second threshold by changing a geohash location in the cube corresponding to the one dynamic geofence zone, and

adjusting the cube by (i) adjusting, based on a model recall accuracy of the one dynamic geofence zone, a number of characters of the geohash code of the cube corresponding to the one dynamic geofence zone to increase or reduce a size of the one dynamic geofence zone in the spatial dimension and (ii) shifting or changing a length of the at least one time period in the temporal dimension;

receiving a request for a trip from a computer device of the rider;

determining that a pickup location is within the one dynamic geofence zone of the at least one set of dynamic geofence zones and that a time of the trip is within a time period of the one dynamic geofence zone of the at least one set of dynamic geofence zones; and

based on the determination, transmitting a request to the computer device of the rider requiring the rider to make a new request from a different location from the pickup location.

16. The non-transitory computer-readable storage medium of claim 15 , wherein updating the plurality of cubes corresponding to the at least one set of dynamic geofence zones further comprises removing at least one dynamic geofence zone from the at least one set of dynamic geofence zones.

17. The non-transitory computer-readable storage medium of claim 15 , wherein adjusting the number of characters of the geohash code of the cube corresponding to the dynamic geofence zone comprises reducing a size of the dynamic geofence zone by increasing the number of characters of the geohash code of the cube corresponding to the dynamic geofence zone.

18. The non-transitory computer-readable storage medium of claim 17 , wherein:

the one dynamic geofence zone of the at least one set of dynamic geofence zones is deployed at a residence address associated with a flagged rider; and

the geohash code of the one dynamic geofence zone of the at least one set of dynamic geofence zones is changed to a new residence address associated with the flagged rider.