Demand prediction now helps ensure that car2gos are available right there where they are really needed. In the future, the precision, and hence the accurate coordination of supply and demand, will be even better. How this will be done and which factors are taken into consideration is explained below.
Well-functioning demand prediction is a precondition for an optimal free-floating carsharing concept, and above all for autonomous carsharing of the future.
The question as to when and where a customer will need a car has to be answered even before the customer actively searches for a vehicle.
What is the purpose of demand prediction?
Why is this? In a competitive mobility market, the customer is most likely to select the provider who offers the best, most reliable service in the shortest time and at a good price – and has exactly the car available at exactly the time that the customer needs.
For the provider this means: the car should ideally already be on its way to the customer even before it has been ordered.
This can be explained more clearly by taking a look into the autonomous future using a practical example.
Practical example: An important soccer match in a large city
Suppose that a soccer match has ended in a stadium in a large city.
There are now two possibilities.
Possibility one: there is no demand prediction. Mobility providers therefore do not know before the end of the game that many fans will be looking for a car after the match.
The fans will walk out of the stadium and order a car with their smartphones.
A large number of cars then have to be driven to the stadium, which takes some time and thus results in long waiting times – waiting times which are inacceptable for the customers.
Possibility two: There is demand prediction, which recognizes that cars will be required at the stadium at the end of the match.
The mobility providers are then able to send cars to the stadium in advance so that they are ready and waiting for the fans as they leave the stadium.
Predicting when and where customers will need cars is already part of the daily business at car2go.
More data – more possibilities
With the data which car2go has collected over the years, the company is able to predict demand extremely accurately using complex algorithms.
External data, such as the weather or dates of events, is also included in the calculation.
It may sound trivial, but when it rains, the demand is higher than it is when the sun shines. And, after a large concert or during a carnival, demand is higher in certain areas of the city than is normally the case.
Demand prediction and data protection – can they work together?
Data protection is also an important topic here. Demand prediction only forecasts that a customer will need a car at a certain time.
It does not determine which customer will be needing a car. Hence, no personal data is collected or stored in this respect.
A look ahead into the future
car2go already constantly predicts the demand in all car2go cities and therefore knows when and where the customers need vehicles.
car2go can thus estimate, for example, how high the demand for vehicles will be next Saturday afternoon in Berlin city center, or on certain streets in New York City.
This helps in the management of the fleet so that the maximum possible availability can be achieved for the customers.
Learning algorithms consistently improve the quality of the prediction through the use of “Advanced Machine Learning Technologies“.
Accurate demand prediction is therefore crucially important for functional autonomous carsharing.
What do you think? Does the somewhat uneasy feeling that a car2go already knows when and where it is needed (unjustly) outweigh the pleasure of optimal availability?