As the digitization of society has bestowed cities with an immense amount of data, cities are using data to inform, monitor, evaluate and measure their performance towards policy objectives. However, it was found that monitoring frameworks are biased by the perspectives of monitor developers, making their viewpoints and opinions dominant in shaping policies and public perception. Next to this, the aggregated metrics of monitoring frameworks overlook the experience of outliers thereby masking and excluding local differences and issues. These limitations lead to a gap between monitoring systems and the reality of lived experiences of citizens. With data-driven decision making being increasingly adopted by policymakers, it is imperative to investigate how these disparities can be minimized.
Literature suggests including citizens in the development of urban monitoring systems can positively contribute to aligning monitoring frameworks with the lived reality of citizens. The goal of this thesis was therefore to explore how citizen engagement with monitoring frameworks can be effectively achieved. In doing so this thesis explored two main components, being 1) how citizens can be effectively engaged with monitoring frameworks and 2) how their local knowledge can be incorporated in monitoring frameworks. This was done through a case study for the Ideal(s) City monitoring framework, developed by the AMS Institute and the City of Amsterdam.
The project took a participatory research-through-design approach where through the testing and evaluation of design interventions practical learnings about citizens’ local knowledge and participative capabilities were combined with theory. Besides, expert interviews and interviews with municipal stakeholders, such as a policymaker and a monitor developer, were conducted to gain a better understanding of the context of urban monitoring and policymaking. By combining the insights of these design interventions, interviews and literature, a process for citizen engagement with accompanying guidelines were developed and the potential role of local knowledge was identified.
It was found that local knowledge of citizens can play three roles in monitoring frameworks: 1) Identifying missing indicators in the current monitoring frameworks to minimize disparities between citizen perspectives and urban monitors, 2) providing new connections among (existing) indicators and 3) assigning weight of importance to indicators to reflect the diverse concerns and experiences of citizens. including this local knowledge can support policymakers in making more informed policy decisions and trade-offs, considering the diverse needs of the local context.
The thesis concludes with presenting a guidebook depicting a process for citizen engagement with monitoring frameworks targeted at monitor developers. Through actionable steps and guidelines, this guidebook aims to support AMS and other monitor developers to set the first steps in aligning urban monitors with the lived experiences of citizens. To ultimately monitor what matters to the city and her citizens.

This graduation project explored the environmental nuisances experienced by houseboat residents in Amsterdam due to vessel traffic. The findings show that the experience of nuisance is influenced by both contextual factors (such as past experiences, type of boat, time, and location) and environmental factors (such as noise and movement).
By reflecting on insights gathered through qualitative research, this project aligns with its initial goal of exploring new methods for measuring environmental nuisance.

Cities want to give birth to their own twins, on a computer.

The urban digital twin is a digital copy of the city constructed from heaps of data rather than concrete, and it is being heralded as the driver for Smart Cities: by collecting more and more data and processing it in more and more sophisticated models we would monitor, predict and control the physical city’s behaviour to engineer solutions for today’s most pressing issues, from climate adaptation to crowd management and from infrastructure to governance.

This research critically examines the role urban digital twins can play in processes of public participation. While regularly mentioned in urban digital twin proposals, little research exists exploring this application, even less that takes a critical stance. One city that is looking to use urban digital twins in participation is Den Haag, which has started work on an urban digital twin called De Digitale Spiegelstad (The Digital Mirror City), and research has taken place primarily in the context of this city and this project.

To this end I iteratively developed visions of what a future urban digital twin for participation could look like. These visions challenged the mainstream or obvious narratives around urban digital twins, following the Adversarial Design philosophy of Carl DiSalvo. This led to a prototype that was used to act out a process of participation concerning the redesign of a playground with residents of the neighbourhood Moerwijk, using the research method of Speculative Enactments developed by Christ Elsden and colleagues. Enactments were followed by group interviews with participants about potential risks and benefits of urban digital twins for participation.

The thesis concludes that urban digital twins may have the potential to make public participation engaging to a wider group of citizens and could contribute to citizen trust and transparency in decision-making, but also poses the risk of steering citizens towards technocratic perspectives and leading conversations to focus on details rather than bigger issues. I provide a series of design recommendations in response to these.

Lastly I reflect on the methods and execution of the project, and the implications this may have for design researchers seeking to embark on a similar journey.

To create safe urban areas, it is important to gain insight into what influences the (perceived) safety of our cities and human settlements. One of the factors that can contribute to safety is the way urban spaces are designed. Previous work has highlighted the importance of natural surveillance: a type of surveillance that is a byproduct of how citizens normally and routinely use the environment. However, studying this concept is not a trivial task. Manual approaches such as observation studies are costly and time consuming and have therefore often limited themselves to smaller geographical areas.

In this work, we present a methodology that can automatically provide an estimate of natural surveillance by detecting building openings (i.e. windows and doors) in street level imagery and localizing them in 3 dimensions. The proposed method is able to estimate natural surveillance at the street segment level, while simultaneously being able to gather data on a whole city in a matter of hours. We then apply our method to the city of Amsterdam to analyze the relationship between natural surveillance and urban safety using the Amsterdam Safety Index.

We conclude that our chosen operationalization of natural surveillance (road surveillability and occupant surveillability) is correlated with decreases in high impact crime and nuisance as well as increases in perceived safety. Furthermore we provide evidence for the existence of a threshold after which extra natural surveillance is no longer associated with higher degrees of safety.

ArenAPoort is an area in Amsterdam’s Zuidoost (Southeast) district, with multiple functions including working, shopping, and entertainment. Most importantly, it is an event area with three famous venues: Johan Cruijff ArenA, Ziggo Dome, and AFAS Live.
Hundreds of events are held in this area every year. During the event period, mobility can always draw a concern. The Operational Mobility Center (OMC) takes charge of the mobility flows, aiming at improving the mobility situation during events in the ArenAPoort. They are now collecting traffic data to help make predictions on traffic situations of future events and to reactively take precautions. This brings the consideration of whether there are other ways of using the data, for example, making use of data to intervene in visitors’ behavior.
In collaboration with the OMC, this master thesis aims to improve the mobility situation around the ArenAPoort during events by making use of data to change visitors’ mobility behavior. The final outcome is an application that can help visitors to plan their event experience in the ArenAPoort.
The project process follows the double-diamond model. It starts with an introduction of the project background, including the context, the organization OMC, and a project brief. After figuring out the background and having the project brief, four research questions are put forward, aiming to have a deeper understanding of the context, gain an empathy with the target group, and seek for theoretical support. Several research activities were taken, including desk research and field research.
According to research results, a design goal is formulated, together with design guidelines. Next, a series of co-creation sessions were performed for ideation. Based on the ideation results, 13 initial ideas are generated, and after two rounds of evaluation, ideas were summarized and integrated.
The next step is to conceptualize the ideas to a complete concept. During this process, the information structure and the user flow of the concept are defined, together with wireframes of key screens. Evaluation is conducted to assess the concept. Based on evaluation results, a final round of iteration was conducted, and the design is finalized.
Six end-users and two experts from the OMC took part in the evaluation of the final design. By summarizing and concluding evaluation results, overall conclusions of the project are drawn, together with the project limitation, future recommendation, and a personal reflection.

Non-recurrent traffic events, consisting of events of an unpredictable nature such as incidents and vehicle breakdowns, can either directly or indirectly influence road traffic. A better understanding of these events could prove beneficial towards improving a multitude of facets concerning the management of the Dutch road network. Traditional traffic event detection, based on significant changes in traffic flow/speed characteristics, is often limited by sparse road sensor coverage. More importantly, traditional detection methods are unable to categorize and describe traffic events.

The aim of this study is to explore to which extent geosocial data (e.g., data from Twitter and Waze) could enrich traditional traffic data (e.g., traffic speed/flow data), in order to improve the detection, categorization, and description of traffic events in the Netherlands. In order to achieve this, a pipeline was designed for extracting knowledge on traffic events from geosocial data sources. We collected geosocial data from Twitter, Waze, and TomTom and used traffic data provided by DiTTLab. We specifically focused on reports by real road users, which we define as natural persons that report on their own account, therefore excluding all legal person entity accounts such as public/private organizations, and bots. A machine learning approach was applied to automatically classify tweets as either traffic event related or not. In order to categorize tweets into a traffic event category, a rule-based traffic domain annotator was created. Additionally, a geocoding method to link tweets to a geographic location was developed. As Waze and TomTom event reports are classified and geocoded by default, we could cluster these reports together with the processed tweets based on their categorical, spatial and temporal extent into a combined traffic event. These combined traffic event reports were then linked to traffic data, based on corresponding spatial and temporal aspects. In order to present the collected data, a web-based interactive map application was built.

This methodology was applied to data collected over the period from 05-12-2017 to 17-02-2018. From the set of collected tweets approximately 6.71% proved traffic event related. Based on a linear support vector machine classification model we achieved an average f1-score of 0.95 and an accuracy of 0.954, for detecting traffic event-related tweets. The rule-based traffic domain annotator showed an average f1-score of 0.874, and an accuracy of 0.964. The geocoding method proved able to geocode tweets to a location that covers all place indicators in a tweet in 86% of the evaluated cases. The remaining 14% of the tweets either got geocoded to a part of relevant indicators or to no relevant indicators at all. Our clustering approach is able to cluster 39.61% of the event reports into a traffic event report cluster consisting out of more than one event report, from which 48.66% could be linked to traffic data.

All in all, based on the achieved results, this work shows that geosocial data can be used to enrich traffic data towards the improvement of the detection, categorization, and description of non-recurrent traffic events.

The digital representations of physical places, known as Points-Of-Interest (POIs), have been the core element of various studies and platforms such as online mapping services (e.g. Google Maps) and location based social networks (e.g. Foursquare). The use of POIs as proxies of the real-world-places facilitates the study of places, urban environments and, consequently, human behavior. Therefore, the extent to which the POIs manage to capture the complex multidimensional nature of physical places defines the limits of all those platforms and of humans' essential understanding of places.

Admittedly, the already existing POI data sources tend to represent differently the physical places (e.g. focus on specific aspects of places) and their data are being produced in a variety of ways (e.g. user generated data or non-user generated data). In addition, multiple sources exist that indirectly include place-related information as, for instance, Google Street View which contains images of the exterior of places without providing a direct link between the image and the corresponding place-entity. Thus, an interesting challenge arises which is how could all those diverse place-related data coming from different data sources be combined towards the creation of a better digital representation of places.

This thesis introduces an innovative approach to the extraction and combination of multidimensional POI features from various place-related data sources towards the study of urban places. It consists of two main parts: (1) the process of selecting, extracting and combining multidimensional POI features from various sources which reflect the high dimensional nature of places and (2) the use of the extracted features to discover which of those - and to what extent - better define and distinguish urban places in respect to their core characteristic, their main function.

Regarding the first part, for the combination of POI data sources a "matching" algorithm is developed whose goal is the identification of POIs which belong to different POI data sources and represent the same physical place and is based on the comparison of a set of attributes such as location, name and website.
For the extraction of the POI features the need of specialized techniques according to the nature of the different data is revealed and several methods are discussed and used.

The second part concentrates in data collected from two capitals, Amsterdam and Athens. A machine learning classifier is trained on different combinations of features extracted from those data and their importance for distinguishing the urban place types is computed and compared. The results, among other, support that the functional (e.g. opening/closing times) and experiential characteristics (e.g. topics extracted from reviews) are the strongest indicators of a place's type independently of the context (e.g. city) while the exterior visual appearance of places does not provide such valuable information. The combination of the extracted features lead to an F1-score of around 60\% when classifying POIs by their type among 10 classes (multiclass problem) and around 90\% when predicting if a POI is of a certain type or not (binary problem).

Overall, the importance of combining multiple data sources in order to capture the complex nature of places is successfully supported by the results and the features that tend to better "describe" places in respect to their main function are discovered and further explored.

Analysing attractiveness of places in a region is beneficial to support urban planning and policy making. However, the attractiveness of a place is a subjective and high-level concept which is difficult to quantify. The existing methods rely on traditional surveys which may require high cost and have low scalability. This thesis attempts to quantify attractiveness of a place in a more efficient way and develop a model which can automatically predict attractiveness based on Street-View data (i.e. from Google Street View).

As a study case, 800 Google Street View images from 200 locations in Amsterdam have been extracted, and their attractiveness perceptions have been evaluated via crowd-sourcing to get the ground-truth information. The other attributes which are presumed to have a relationship with attractiveness are also assessed, such as familiarity, uniqueness, friendliness, pleasure, arousal, and dominance. The research and analysis revealed several insights related to the attractiveness of places. Attractive perception when seeing a place is positively correlated with perception of uniqueness, friendliness, pleasure, and dominance. Moreover, pleasure is possibly multi-collinear with attractiveness. It was also found that attractiveness perception has low spatial auto-correlation, which means that nearby places do not necessarily have similar attractiveness. Some visual features related to attractiveness were also investigated. The result indicated that scenes related to roads and residential buildings are less attractive, meanwhile, scenes related to greenery, blue sky, and water environment are more attractive.

A Convolutional Neural Network (CNN) model has been developed via machine learning approach which could automatically predict attractiveness perception of a place based on its representing Google Street View image. The developed model achieved 55.9% accuracy and RMSE of 0.70 to predict attractiveness in 5 ordinal values.