Weather forecast and Google reviews make car sharing more profitable
Car-sharing rental companies and companies like Uber can be more profitable by anticipating peak times that occur at popular spots in the city or as a result of bad weather. This is what Micha Kahlen, PhD candidate at Rotterdam School of Management, Erasmus University (RSM), discovered. His model can help such companies deploy the vehicles better throughout a city, resulting in customers being helped quicker and fewer cars being deployed.
Companies like Car2Go and even Uber are still leaving it largely to chance where the cars are precisely located for renting or driving looking for potential customers, says Kahlen. In the case of Uber, often the nearest car is sent to the customer without anticipating the expected demand. Car-sharing rental companies also often fail to make conscious decisions about where they place their cars and, as a result, they can lose customers.
To find out how to improve predicting the demand for cars, Kahlen studied the GPS information and the journeys made by 2,400 cars of Car2Go and DriveNow around Berlin over a five-month period. He compared this data with the weather forecasts on these days and the points of interest in Google Maps, because when it rains, people prefer to carshare.
The researcher looked at how many of these vehicles were in the vicinity of points of interest on Google Maps – such as train stations and department stores – and how many stars they were rated. He expected greater demand around popular spots.
He discovered that at certain times, the cars were not in the right places to meet the specific demand. He used the data to design a model that can predict in real time the number of cars needed in a certain zone for every hour in all parts of the city. The results are based on car-sharing rentals, but Kahlen believes the model can be applied to all companies offering similar mobility services.
His calculations showed that companies can use this model to increase their profits, even after the relocation of the cars has been included. The precise amount of profit depends on the type of company: for Car2Go, this is three per cent. If a company like Uber were to use the model to deploy its cars, profits could rise by six per cent, or even to seven per cent by switching to autonomous vehicles.
To achieve this, the model also showed that companies will need to reduce the number of cars they deploy by 23 to 31 per cent. The question is whether this strategy is so sensible in the long term. Just driving around popular places and at specific times would probably make the company less popular among some of its customers, concludes Kahlen.
Download Micha Kahlen's PhD thesis here (after the PhD defence of 15 September 2017).