A DSS to Improve the Performance of Car-Sharing Schemes

Car-sharing is an innovative and sustainable mobility concept. It represents a potential solution for cities that try to take measures to reduce the wasteful use of cars and their environmental impact, marking an important shift from vehicle ownership to service use (Prettenthaler and Steininger 1999). Having become popular worldwide in the 1990s, the overall car-sharing service has grown rapidly in recent years, reaching several million customers and counting a fleet of some tens of thousands of vehicles (Le Vine 2014). Furthermore, consolidated projects on car-sharing services have been implemented in more than 600 cities in over 18 countries (Shaheen and Cohen 2013). The car-sharing market also witnessed a growing interest in delivering the service from various types of organizations, such as for profit (Greenwheels 2016; Communauto 2016) and not-for-profit operators (Carma 2016), public transport operators (Flinkster 2016; Transdev 2016), car manufacturers (Daimler 2016; BMW 2016), and citizen networkers (Getaround 2016; Drivy 2016; Tamyca 2016).

As more players enter this new market, the competition among them is intensifying. Along with the strong interest in this innovative type of mobility, all car-sharing players have to be prepared to face the challenges and threats of this new market.

A clear need thus arises for a valuable decision-supporting tool, able to help car-sharing providers develop efficient strategic and operational planning as well as continuously monitor the performance of the service.

Starting from this context analysis, we have developed a DSS that is specifically designed to be a valuable solution to support the strategic decisions of the operators and improve their business performance.

The DSS was developed thanks to strong cooperation by private companies that operate in the green economy sector and provide innovative and environmentally sustainable car-sharing services in Italy. The cooperation with real players allowed us to build the DSS using real data (about 6000 clients and 2 years of rentals) to design a DSS that easily adapts to different car-sharing services (i.e., station based or free-floating; traditional or EVs). The DSS is a web-based tool that supports the car-sharing provider with specific infographics based on historical data showing the evolution of the car-sharing service. A simple dynamic graphic interface allows easier understanding of complex data and information. In addition to the representation of historical data related to the service performance, the DSS provides mathematical models for demand forecasting, operational decisions, and service optimization. The DSS is composed of three sections that reflect the demand and supply characteristics of car-sharing systems: (1) the client section, (2) the service section, and (3) the relocation section.

The first section collects and analyzes client data sets and displays a series of information through different infographics. This information could be relevant for developing marketing strategies and implementing better planning of the service. Splitting by age and by number of car rentals made in a given period for all users registered to the service, it was possible to apply an appropriate clustering algorithm that creates distinct groups of clients. Through a reliable cluster analysis, the DSS could in fact help the car-sharing provider to differentiate the profitable clients from the nonprofitable ones. Consequently, the provider could develop a more focused marketing campaign customized to each of the identified customer groups. As an example, Figure 14.2 shows the result of a clustering analysis, performed by means of the k-mean algorithm. The analysis of the data set clearly identifies four classes of users.

Furthermore, this first section also collects and maps data related to the geographic distribution of clients and information related to the popularity level of each zone of the operational area of the car-sharing service. In fact, in order to develop an accurate demand model, it is necessary to discretize the operational area in zones comparable with micro geographic units with homogeneous socioeconomic characteristics, since the dislocation of activities and services within the operational area influences demand distribution. Through the visualization of these data, DSS users could have instant feedback about the existing spatial distribution of clients. This information could be used to find the optimal distribution of car-sharing vehicles within a given zone, matching car-sharing demand and supply. More specifically, by taking into account the starting point of the customers and their preferences about departure and arrival zones, the car-sharing provider could take important decisions in terms of strengthening or weakening existing zones and extending or restricting the operational area.

The second section gives the user an at-a-glance overview of the performance of the carsharing service. All the data are presented in easy-to-understand histograms. First, it gives information about the number of car rentals in a given interval of time since the beginning of the service. Second, as shown in Figure 14.3, it provides an overview of the service usage data with specific statistical measures on a zone-by-zone basis. These measures are related to the number of arrivals; the number of departures; the workload number, which identifies the most-used zones since it is calculated as the sum of departures and arrivals in each zone; and the car needs number. This last information allows the car-sharing provider to understand if a zone is used mainly as a starting point and thus potentially needs

FIGURE 14.2

Result of the clustering analysis; points are the clients, with client age along the x-axis and total number of rentals of each client along the у-axis; the circles are the representative cluster users.

FIGURE 14.3

Performance dashboard. The x-axis defines the zone identification number.

a greater number of vehicles, or alternatively if it is used as a point of arrival and therefore already has vehicles. In other words, it could indicate a potential shortage or accumulation of vehicles at certain zones. In this way, it gives, through a simple view, a first input to the planning of its relocation strategy.

Built to be regularly updated (e.g., weekly or monthly), the service section is a sort of performance dashboard that displays the most important information to decision makers so that they can monitor and manage the service more efficiently.

The first two sections were developed as data analysis systems (Power 2002), which means that they collect, analyze, and display in an intuitive manner all current or historical data about the clients and the service performance. Furthermore, the use of a dynamic graphic interface allows users to query data in real time, evaluate the performance of the analyzed system, and consequently make decisions. The relocation section goes a step further than the first two sections. It is an extension of the data analysis system since it processes the data collected using a forecast model for the prediction of future demand. The result of this modeling technique is an intuitive tool that supports the decision-making process related to relocation strategies. Relocation is defined as an activity that is performed by the car-sharing operator staff to prevent or minimize vehicle imbalance issues, moving vehicles between zones having vehicles in excess and zones having a lack of vehicles. In order to efficiently rebalance the resources of the service, it would be useful for the car-sharing operators to know the number of vehicles required to meet the demand for each of the most-stressed zones. For this purpose, we have implemented an algorithm in our tool, able to predict this number by using different historical data sets related to the service usage of each zone. Setting a specific time slot (hour and day of the week), the tool gives information about the probability of a zone receiving a request for a rental. This leads to the implementation of efficient relocation strategies that maximize the number of satisfied requests and reduce logistics costs.

This DSS represents a powerful tool, generated by the fruitful collaboration between the business and research communities. The first one made the data sets available in order to optimize its overall performance in a more efficient way, while the second one provided the scientific methodologies to analyze and visualize the data sets to have a better understanding of the business and operational aspects and support the diffusion of this new type of sustainable mobility.

If the current DSS supports the decision making of an operator running the service, one of the main goals of our future work is the development of a tool that supports the service design process of a car-sharing company that wants to enter a new city, both in the case of the absence or coexistence of other car-sharing providers in the same environment. The tool will aim to determine the fleet size, which is an important asset in order to meet customer demand and optimize the service within the covered area, and to choose the service model that best suits the environment in which it will operate. In addition, it will be a useful tool for the evaluation of different service-planning alternatives on the basis of cost structures and revenue streams.

 
Source
< Prev   CONTENTS   Source   Next >