In 2017, the Swiss Federal Railways (SBB CFF FFS) launched the SBB Green Class[1] project, a one-year pilot study in which 140 participants across Switzerland were offered a comprehensive Mobility-as-a-Service (MaaS) package, containing (among others) a general public transport pass valid in the entire country (GA) and an electric car (BMW i3). In order to assess the customers’ reactions to this unprecedented offer, the participants were tracked with a smartphone application for the entire project duration. The resulting data were processed and analyzed by the Mobility Information Engineering (MIE) Lab[2] at the Chair of Geoinformation Engineering[3], ETH Zurich, with a particular focus on assessing the potential of such mobility packages for inducing more sustainable transportation behavior.

Figure 1 illustrates the chosen framework for movement data analysis. In this case, movement data were obtained from two sources, i.e., the smartphone application and the electric car used by each participant. The raw trajectory data were then converted to a more suitable format and imported into a spatial database. Before the analysis, various preprocessing steps were necessary, including imputing missing values and records, matching the movements to the transportation network, checking the plausibility and quality of the data, as well as filtering for anomalies. All of these steps relied on machine learning and data mining methods, which were also utilized for the subsequent data analysis. Here, the focus was placed on creating both detailed user profiles (e.g., detecting frequently visited places or habitual travel patterns) and aggregated mobility metrics (e.g., assessing the longitudinal development of CO2 emissions or the modal split).

Figure 1: Analysis framework

At the present stage, movement data from the first 7 months (February – August 2017) have been analyzed. Figure 2 illustrates the development of the average CO2 emissions for all SBB Green Class users per calendar week and mode of transport, including 6 weeks of movement data recorded prior to the project start. A clear drop in emissions is visible with the start of the project in February 2017 (KW 4). This is accompanied by a drop in terms of the relative share of CO2 emissions of the conventional car, which becomes clearer when analyzing the modal split, as seen in figure 3.

Figure 2: Average CO2 emissions for all users

Here, after the project start, one can see the relative share of conventional car trips decreasing drastically with the emerging usage of the electric car. The modal split for the other modes of transport remains relatively static. What is also striking is an increase of conventional car usage in the summer months, potentially due to the holiday period.

Figure 3: Modal split for all users

The trend of the electric car replacing the conventional car to some degree is also visible in figure 4, which illustrates the partial correlation matrix for all modes of transport. In the highlighted part, a strong negative correlation can be seen between the usage of the conventional car and the electric alternative, meaning that with decreasing distances of the former come increasing distances traveled with the latter mode of transport.

Figure 4: Partial correlation matrix for all modes of transport and users

In conclusion, the findings demonstrate that the SBB Green Class offer indeed resulted in a relative decrease of CO2 emissions produced by its customers, which appears to be mainly caused by a partial replacement of the conventional car by the electric car. Furthermore, the results demonstrate the practical potential of GPS-tracking as data source for large-scale, longitudinal mobility studies.


The people involved in this project are Dr. David Jonietz, Henry MartinDominik Bucher, and Prof. Martin Raubal.

Professor Raubal is an ESC member and Professor of Geoinformation Engineering at the Institute of Cartography and Geoinformation.

The Chair of Geoinformation Engineering has its research foci in the areas of mobility and energy. In the area of mobility, the group research lies at the intersection of mobile geographic information systems, geospatial information technologies, and mobile decision-making. In the area of energy, the focus is on the development of spatio-temporal algorithms to determine the technical and economic potential of renewable energy sources, and optimal locations for energy facilities and power lines.