Data-Driven City Planning

Managing urban mobility during peak tourist periods requires proactive, forward-looking city planning. To resolve recurring bicycle parking shortages along the Scheveningen beach coastline, the Smart Sensor Systems (SSS) research group, part of Centre of Expertise Digital Operations & Finance, of The Hague University of Applied Sciences (THUAS) partnered with the Municipality of The Hague. Through a dedicated research project supervised by Amey Vasulkar at the SSS research group and executed by Master Next Level Engineering (NLE) student Joost Bleeker, the collaboration delivered an interpretable forecasting model engineered for live integration into the municipality's Crowd Safety Manager (CSM) Digitwin.

The Coastal Capacity Challenge

During warm summer weekends, the Scheveningen boulevard attracts up to 180,000 daily visitors, with approximately 22% arriving by bicycle. This rapid influx regularly overwhelms designated guarded bicycle parking facilities. Bicycles subsequently parked outside designated zones obstruct pedestrian walkways and vital emergency service routes. Until this project, the municipality's deployment of temporary pop-up parking facilities relied on reactive field observations rather than predictive capacity planning. 

Our Role and Research Approach

The Smart Sensor Systems research group contributed data science expertise and academic supervision to bridge the gap between the municipality's urban mobility data and their operational decision-making.

Early data quality analysis revealed a critical technical obstacle: historical occupancy logs from existing Biesieklette guarded facilities suffered from severe data drift and unrecorded exits, rendering them unsuitable as a reliable machine learning target. To solve this, a methodological pivot: utilizing automated induction loop counts from traffic control system (VRI) was used as a reliable proxy indicator for incoming bicycle flow. 

The research compared predictive models, specifically SARIMAX and Prophet, utilizing historical mobility data alongside meteorological weather data, holiday and municipal event calendars. 

Key Findings and Project Outcomes

Prophet was selected as the definitive model due to its superior performance in handling overlapping daily, weekly, and yearly seasonality’s. The research yielded several core operational assets:

• Predictive Benchmark: Evaluated on a held-out 2025 test year, the model achieved a Mean Absolute Percentage Error (MAPE) of 19%, successfully outperforming a seasonal naive baseline.
• Identified Demand Drivers: The model confirmed that global radiation and air temperature are the strongest positive drivers of recreational cycling demand, while wind gusts and relative humidity negatively impact bicycle flow. 
• Actionable Dashboard Integration: Through a User-Centred Design (UCD) approach, a functional dashboard prototype was developed for municipal staff.
• Scenario Alignment: Rather than presenting abstract vehicle counts, the dashboard converts predictions into historical percentile ranks and links them directly to the municipality's established green, yellow, and orange crowd management scenarios.

Field Validation, Long-Term Municipal Adoption & Future Collaboration

The practical viability of the system was proven during the dashboard's user testing phase. The model flagged an unusually high demand forecast for an upcoming Friday; municipal staff trusted the data signal and pre-emptively activated a higher crowd management scenario, deploying extra traffic controllers and field personnel to the boulevard. Actual field measurements that Friday confirmed the surge, validating the model's predictive accuracy. 

Following the successful execution of the research, Joost Bleeker defended his thesis for the NLE Master program, being awarded an exceptional final grade of 9. Recognizing the immediate operational value of the project, the Municipality of The Hague has confirmed its intent to deploy the system within their live CSM Digitwin framework. Furthermore, the municipality plans to establish an ongoing collaboration pipeline with the SSS research group, utilizing future THUAS students to solve complex urban mobility challenges.

For more information about this research, contact Amey Vasulkar, [email protected]