Taxi ride scheduling and pricing using historical data

Abstract

Taxis often bring people to airports and drive back to their area of operation without a passenger. This is considered as a gap in a taxi driver's schedule and could be filled up by giving customers extra incentive to book a taxi ride by asking a reduced price. A reduced price is justified when the costs of when the rides are combined is lower than when the rides are carried out separately. This thesis work describes two steps towards computing the cost of performing a taxi ride while taking into account the time and fuel needed for driving empty between rides. The first method responds to a request of a customer and composes offers that are related to the customer's taxi ride request from different companies, while taking the already booked rides into account. A good offer has both the price and the offset to the requested departure time minimized. The second method learns the cost for driving empty by taking into account the probability of taxi rides that will be booked in the future.