Predictive AlgorithmBoetho’s Bayesian Machine Learning algorithm predicts future parking occupancy based on current payment data. Recognizes patterns, trends and seasonality.
Boetho's algorithm takes in as input past measurements of parking occupancies and returns as output predictions of future occupancies for the upcoming days. The amount of data that is taken as input, as well as the number of days of forecast that is produced can vary, with the precision of the prediction being proportional to the amount of past measurements and inversely proportional to the amount of future predictions required. Considering a normal scenario (for many interesting applications) of generating a week ahead forecast with hourly data points, in order to obtain an accurate prediction, two months of past data are required. In such a case, Boetho's algorithm can produce accurate predictions even considering payment data as a proxy, thus not needing expensive sensors or cameras to operate. The algorithm can be equally applied to off-street parking locations as well as on-street parkings, where each spot can be treated individually or different zones of the city can be considered as a whole. |
Smart PricingOur Smart Pricing formula updates parking fees according to predicted occupancy rates, matching prices to demand. Fees fluctuate in a clear structure for ease of customer understanding.
When yield management was first introduced in the airline business, it completely revolutionised that industry. Revenue management has experienced a similar success in other areas, such as e-commerce and the hospitality sector. Thanks to Boetho's predictive algorithm, an optimization of parking prices, based on the forecast of clients' behaviour, can now be implemented in the parking industry as well. By increasing the price of parking spots during the hours of the day when high occupancy is expected and decreasing them when low occupancy is anticipated, an increase in revenues can be achieved. Moreover, since Boetho's algorithm can work on the readily available payment data, this optimization of the revenue stream can be easily achieved, without the need for costly initial investments, such as the deployments of sensors and cameras. |
Optimal AllocationIn large scale applications, our Optimal Allocation algorithm ensures that a global optimal use of the parking resources is reached.
When dealing with large scale applications of our predictive algorithm, the following aspect must be taken into account: the behaviour of drivers will inevitably change if information of future parking availability is presented to them or if new strategies are implemented by the parking owners. This advocates for the necessity of adaptive algorithms, that are able to capture new and unprecedented trends in the data, as well as optimization algorithms that can take into account the reaction in the drivers’ behaviour. That is why at Boetho we have developed an optimal allocation algorithm that by suggesting different actions to different drivers, is able to obtain an optimal use of the parking resources of a city or network of parking. Hence providing a solution for the great deal of parking related problems that afflict a city, such as traffic jams, pollution, time wasted by the drivers in search of a free parking spot. |
Booking System OptimisationParking owners can optimize the revenue stream from their booking systems by matching prices with future demand, as well as minimize the time needed to keep a spot reserved.
Any off-street parking location can implement a service that will allow its customers to book a parking spot for a certain future day and time. In practice, not many do so, due to the managerial problems related to this type of service. Indeed, it is not clear for how long a spot is to be kept free, in order to guarantee a free space when a customer with a reservation arrives and consequently how much to charge for it. The option of leaving a spot free all day for every booking is, of course, available, but not efficient and the high cost of the service may discourage a multitude of potential clients. However, the knowledge of drivers' future behaviour provides the perfect piece of information needed to make this booking system a reality. By predicting the number of customers leaving every hour or half an hour, one can easily identify the amount of time that a parking space has to be kept free, before the booking time, and hence provide a competitive (and optimized) price for the service, all thanks to Boetho's predictive algorithm. |
If the only constant is change then we will keep looking for better ways.
|