In many places, particularly low cycling maturity cities, several investments in cycling infrastructure are made without considering the overall network impact. One of the most frequent examples is placing a piece of network segment that does not connect to other cycling network parts, despite research showing that a cycle network retrieves more modal shift gains if well connected.
The result of subsequent such practices is a cycling network that is made of several segments that do not join together anywhere instead of a proper well-connected network, which results in users not being able to use the bicycle fully as a means of transportation.
Planning the infrastructure needed to change mobility patterns requires high-quality evidence. Data and models are needed to ensure cost-effective investment, enabling new infrastructure to be planned where most needed. In the case of cycling, it has the greatest potential to replace unnecessary and unhealthy short motorized trips in more dense urban centers, which means designing efficient cycleways and safe street connections.
Create a data-driven framework/solution for finding the missing segments in a scattered cycling network of a city.
GOAL 11: Sustainable Cities and Communities
The following datasets were provided to the participants:
Besides the provided data, the data provider, suggested other data sources, such as the UK National Travel Survey and the Sydney Travel Survey and cycling count data, which has information about the cycling habits of people in the UK and Sydney.
Teams also used open data from GIRA, the shared bicycle system in place in Lisbon, and information about the location of public transportation, such as trains and metro, in which a bicycle could be transported.
Due to the nature of this challenge, all teams focused extensively on data analysis and exploration. Some techniques used were merging different data sources, data normalization, and plotting information in map visualizations.
One team computed distance centroids per district of Lisbon to identify locations where there is a lack of bike lanes close to those district centroids. They handled data from 24 Lisbon districts provided with 3502 bike lanes. In addition, they analyzed the mobility patterns of each district to find relevant discrepancies based on data from other means of transportation, such as car, metro, and train. They also interpreted the problem as a Graph Network analysis issue and therefore built a graph network to analyze the interconnectivity between districts, from which they derived several conclusions.
One team used data from several means of transportation to understand where citizens usually want to go as a way to identify places where cycle networks should exist. On top of that, they used external data from the shared bicycle system of Lisbon, which, coupled with bike usage data, enabled them to understand where new bike segments should go. With that in mind, they identified the districts with the highest mobility (from all modes of transportation) and highest relative bike usage. Using that information, the team identified the districts with the most unnecessary car trips (because of their shortness) that could easily be done by bike instead.
Lastly, this team also produced the visualization in Figure 1, which gives an overall idea of the cycling network in Lisbon and how connected it is with public transportation. It also shows the inclination of roads, which was an additional factor that this team considered when suggesting new segments of bike pathways.
According to this team, the main missing link is connecting the current cycle network that ends at Avenida da Liberdade with Terreiro do Paço through Baixa-Chiado. This segment is crucial, considering the inclination of the road (completely flat), the presence of bike-sharing stations, the connectivity to public transportation (bus, metro, and ferry), and the mobility patterns of citizens. Additionally, in their work, they detail more suggested segments and the specific districts and areas of Lisbon.
Another team focused on connectivity between districts by analyzing the number of cycling roads that connect each one of them. Some districts are very well connected, both intra-district and inter-district, and others lack inter or intra-connectivity. For example, the districts of Ajuda and Campo de Ourique are the only two districts completely disconnected via bike from their neighbors, and for that reason, they could be good candidates for connection.
This team also suggested that a good criterion for connecting two cycling lanes would be to consider the shortest distance, meaning that current bike lanes that are closest to each other should be connected. Alternatively, by calculating the centroid of mobility of each district, another criterion could be the shortest distance to that centroid.
