Methods and servers for generating a prediction score by a machine learning algorithm
A method and server for generating a prediction score by a Machine Learned Algorithm (MLA). The method includes: acquiring route data, generating a plurality of route scores for the plurality of routes based on the route data, acquiring driver data, generating a plurality of driver scores for the plurality of drivers based on the driver data, and for a given route-driver pair: generating by the MLA a given prediction score indicative of a likelihood of the given driver having an accident on the given route within a predetermined amount of time after initiating the given route. The method also includes assigning, by employing a dispatching algorithm based on the prediction score, the given route to the given driver for providing a transportation service to the user.
1. A method of generating a prediction score by a Machine Learned Algorithm (MLA), the MLA being executed by a server, the server executing an digital transportation service, the server being configured to be communicatively coupled to a plurality of electronic devices associated with respective users and respective drivers of the digital transportation service, the method executable by the server, the method comprising:
acquiring, by the server, route data associated with a plurality of routes requested by users of the digital transportation service;
generating, by the server, a plurality of route scores for the plurality of routes based on the route data, a given route score being indicative of how safe a respective route from the plurality of routes is;
acquiring, by the server, driver data associated with a plurality of drivers of the digital transportation service;
generating, by the server, a plurality of driver scores for the plurality of drivers based on the driver data, a given driver score being indicative of a driving behavior of the respective driver from the plurality of drivers;
for a given pair including the given route and the given driver:
generating, by the server employing the MLA, a given prediction score indicative of a likelihood of the given driver having an accident on the given route within a predetermined amount of time after initiating the given route;
assigning, by the server employing a dispatching algorithm, the given route to the given driver for providing a transportation service to the user, the assigning being based on the prediction score.
2. The method of claim 1 , wherein the generating the route score comprises generating, by the server, a ratio of accidents on the route over a route length of the route.
3. The method of claim 1 , wherein the generating the driver score comprises generating, by the server employing a driver model, the driver score based on driver behavioral data and driver profile data.
4. The method of claim 1 , wherein the method further comprises generating, by the server, a dispatch time value for the given pair, and wherein the assigning is further based on the dispatch time value.
5. The method of claim 1 , wherein the MLA is a logistic regression model.
6. The method of claim 1 , wherein the MLA is a pairwise ranking model.
7. The method of claim 1 , wherein the dispatching algorithm is a Hungarian algorithm.
8. A method of dispatching a driver to a ride request, the method executable by a server, the server executing an digital transportation service, the server being configured to be communicatively coupled to a plurality of electronic devices associated with respective users and respective drivers of the digital transportation service, the method comprising:
acquiring, by the server, route data associated with a plurality of routes requested by users of the digital transportation service, a given route from the plurality of routes being associated with a respective ride request;
generating, by the server, a plurality of route scores for the plurality of routes based on the route data, the given route score being indicative of how safe a respective route from the plurality of routes is;
acquiring, by the server, driver data associated with a plurality of drivers of the digital transportation service;
generating, by the server, a plurality of driver scores for the plurality of drivers based on the driver data, a given driver score being indicative of a driving behavior of the respective driver from the plurality of drivers;
for a given pair including the given route and the given driver:
generating, by the server employing the MLA, a given prediction score indicative of a likelihood of the given driver having an accident on the given route within a predetermined amount of time after initiating the given route;
assigning, by the server employing a dispatching algorithm, the given route to the given driver for providing a transportation service to the user, the assigning being based on the prediction score; and
transmitting, by the server, a request for confirmation to the given driver for accepting the given route.
9. A server for generating a prediction score by a Machine Learned Algorithm (MLA), the MLA being executed by the server, the server executing an digital transportation service, the server being configured to be communicatively coupled to a plurality of electronic devices associated with respective users and respective drivers of the digital transportation service, the server being configured to:
acquire route data associated with a plurality of routes requested by users of the digital transportation service;
generate a plurality of route scores for the plurality of routes based on the route data, a given route score being indicative of how safe a respective route from the plurality of routes is;
acquire driver data associated with a plurality of drivers of the digital transportation service;
generate a plurality of driver scores for the plurality of drivers based on the driver data, a given driver score being indicative of a driving behavior of the respective driver from the plurality of drivers;
for a given pair including the given route and the given driver:
generate, by employing the MLA, a given prediction score indicative of a likelihood of the given driver having an accident on the given route within a predetermined amount of time after initiating the given route;
assign, by employing a dispatching algorithm, the given route to the given driver for providing a transportation service to the user, the assigning being based on the prediction score.
10. The server of claim 9 , wherein to generate the route score comprises the server configured to generate a ratio of accidents on the route over a route length of the route.
11. The server of claim 9 , wherein to generate the driver score comprises the server configured to generate, by employing a driver model, the driver score based on driver behavioral data and driver profile data.
12. The server of claim 9 , wherein the server is further configured to generate a dispatch time value for the given pair, and wherein the assigning is further based on the dispatch time value.
13. The server of claim 9 , wherein the MLA is a logistic regression model.
14. The server of claim 9 , wherein the MLA is a pairwise ranking model.
15. The server of claim 9 , wherein the dispatching algorithm is a Hungarian algorithm.