Method and system for predicting demand for vehicles
The disclosed embodiments illustrate methods and systems for predicting demand of vehicles in a transportation network. The method includes determining demand events at each of one or more locations for time intervals based on historical demand data. The demand events correspond to a demand of vehicles at one or more locations during plurality of time intervals throughout a day. The method includes creating a graph comprising nodes, and edges connecting nodes, each node being representative of a demand event from demand events. An edge is representative of dependency between two demand events from demand events. The method includes predicting demand of vehicles at a location from one or more locations during a predetermined time interval based on the graph and a real time demand of vehicles associated with other demand event. The method includes displaying demand prediction on a computing device at one or more locations of transportation network.
1. A method for deploying one or more public transit vehicles in a public transit network, the method comprising:
determining, by one or more microprocessors, historical demand values associated with a plurality of demand events based on historical ticket sale data of a plurality of public transit nodes during a plurality of time intervals,
wherein each of the plurality of demand events is associated with a public transit node of the plurality of public transit nodes and a time interval of the plurality of time intervals, and
each of the historical demand values is associated with one of the plurality of demand events and equals the total number of public transit vehicle tickets sold at the public transit node associated with the demand event during the time interval associated with the demand event:
determining, by the one or more microprocessors, dependencies between the plurality of demand events based on the determined historical demand values, each of the dependencies quantifying a correlation between a demand of one of the plurality of demand events and a demand of another one of the plurality of demand events;
creating, by the one or more microprocessors, a graph comprising graph nodes and one or more graph edges,
wherein each of the graph nodes is associated with a demand event of the plurality of demand events, and
wherein each graph edge of the one or more graph edges connects two of the graph nodes;
assigning, by the one or more microprocessors, weights to the one or more graph edges of the graph based on the determined dependencies between the plurality of demand events,
wherein, for each graph edge of the one or more graph edges, the weight assigned to the graph edge corresponds to a strength of a dependency between the demand events associated with the two graph nodes connected by the graph edge;
predicting, by the one or more microprocessors, a future demand for public transit vehicles at a given public transit node of the plurality of public transit nodes during a given time interval of the plurality of time intervals based on the created graph and a real time ticket sale data indicating a quantity of tickets sold at another public transit node of the plurality of public transit nodes;
deploying, by the one or more microprocessors, a number of the public transit vehicles at the given public transit node that will meet the predicted demand; and
wherein determining the dependencies between the plurality of demand events based on the determined historical demand values comprises determining, by the one or more microprocessors, the dependencies using a linear regression model.
2. The method of claim 1 , wherein the linear regression model comprises at least a Poisson distribution, lasso regression model, or penalized lasso regression.
3. The method of claim 1 , wherein the graph corresponds to at least a direct acyclic graph.
4. The method of claim 1 , wherein predicting the future demand comprises predicting the future demand based on the weights assigned to the one or more graph edges in the graph.
5. A system for deploying one or more vehicles in a transportation network, the system comprising:
one or more microprocessors configured to:
determine historical demand values associated with a plurality of demand events based on historical ticket sale data of each of a plurality of public transit nodes during a plurality of time intervals,
wherein each of the plurality of demand events is associated with a public transit node of the plurality of public transit nodes and a time interval of the plurality of time intervals, and
each of the historical demand values is associated with one of the plurality of demand events and equals the total number of public transit vehicle tickets sold at the public transit node associated with the demand event during the time interval associated with the demand event;
determine dependencies between the plurality of demand events based on the determined historical demand values, each of the dependencies quantifying a correlation between a demand of one of the plurality of demand events and a demand of another one of the plurality of demand events;
create a graph comprising graph nodes and one or more graph edges,
wherein each of the graph nodes is associated with a demand event of the plurality of demand events, and
wherein each graph edge of the one or more graph edges connects two of the graph nodes;
assign weights to the one or more graph edges of the graph based on the determined dependencies between the plurality of demand events,
wherein, for each graph edge of the one or more graph edges, the weight assigned to the graph edge corresponds to a strength of a dependency between the demand events associated with the two graph nodes connected by the graph edge;
predict a future demand for public transit vehicles at a given public transit node of the plurality of public transit nodes during a given time interval of the plurality of time intervals based on the created graph and a real time ticket sale data indicating a quantity of tickets sold at another public transit node of the plurality of public transit nodes;
deploy a number of the public transit vehicles at the given public transit node that will meet the predicted demand; and
wherein the one or more microprocessors are further configured to determine the dependencies using at least a linear regression model.
6. The system of claim 5 , wherein the linear regression model comprises at least a Poisson distribution, lasso regression model, or penalized lasso regression.
7. The system of claim 5 , wherein the graph corresponds to at least a direct acyclic graph.
8. The system of claim 5 , wherein the demand is predicted based on the weights assigned to each of the one or more graph edges in the graph.
9. The system of claim 8 , wherein the one or more microprocessors are further configured to instruct the vehicles to be deployed at the particular given public transit node based on the predicted demand.