Congratulations to all four winners of the SLA Student Awards 2025. The standard this year was once again of a very high standard and our judges had a hard time choosing between them. Overall we had over 20 entries from four different universities:
- University College London
- University of Leeds
- University of Greenwich
- University of Bristol
Thanks to everyone for sharing and encouraging students to enter.
SLA Student Awards winners
We awarded prizes in the Masters and Undergraduate categories:
SLA Student Awards – Masters
- Chung-En Tsern (UCL) – Dynamic Urban Function Representation: Learning Urban Characteristics from POI, Land Use and Human Mobility Patterns
- Naieme Golzan Osguie (Leeds) – Through a Machine’s Lens: What Deprivation Looks Like in Leeds
SLA Student Awards – Undergraduate
- Conor Nugent (Leeds) – Stranded: A Methodological Framework for Quantifying the Risk of Loneliness in a Digital World
- Saisruthi Dinesh (Greenwich) – Urban Analytics for Fitness Infrastructure: Identifying Optimal Gym Locations in London Using Multi-Criteria Decision Analysis and Spatial Intelligence
Meet our Winners
First Prize – Masters
Chung-En Tsern – University College London
Biography – Chung-En Tsern is from Taiwan with a background in data science and design. He focuses on data-driven approaches to understanding and shaping cities, combining spatial analysis, quantitative modelling, and AI. His interdisciplinary work connects design thinking and computational tools to support decision-making across government and industry. He recently completed his MSc in Urban Spatial Science at UCL CASA and hopes to continue exploring data and AI applications for urban innovation in the UK.
Dissertation Title: Dynamic Urban Function Representation: Learning Urban Characteristics from POI, Land Use and Human Mobility Patterns
Synopsis of Dissertation:
Conventional land-use and census classifications provide essential baselines for urban analysis but remain static, slow to update, and unable to reflect the temporal rhythms or socio-demographic heterogeneity of contemporary cities. Advances in geospatial representation learning highlight the potential of combining Points of Interest (POIs), mobility data, and embedding techniques, yet most existing approaches either treat POIs as static features or employ spatiotemporal signals mainly for forecasting tasks with limited interpretability.
This study develops a spatiotemporal embedding framework to delineate urban functions in Greater London by integrating POI distributions, land-use proportions, and anonymised mobile phone mobility data, aggregated into H3 hexagonal grids. Spatial semantics are represented through compact numerical embeddings of POIs and land use, while temporal dynamics are modelled using transformer-based attention mechanisms to capture day–night cycles, and demographic patterns. These embeddings are fused into a unified representation that uncovers coherent functional communities, aligns with official land-use and census-based classifications, and reveals urban characteristics that static datasets alone cannot capture.
To evaluate this framework, three approaches to delineating London’s urban functions are compared: (1) conventional land-use classifications, (2) the census-based London Output Area, and (3) dynamic clusters derived from fused mobility, POI, and land-use embeddings. By analysing the overlaps and divergences between these approaches, the study quantifies how mobility patterns and demographic profiles reshape London’s functional landscape.
The research focuses three key questions: (1) To what extent can joint embeddings of POIs, land use, and mobility data capture the city’s functional structure? (2) Can model detect short-term temporal variations such as day–night and weekday–weekend cycles? (3) Can the model differentiate socio-economic stratification across urban zones by learning demographic profiles from mobility data?
Second Prize – Masters
Naieme Golzan Osguie – University of Leeds
Biography: Bridging her background in urban planning with a growing expertise in data science, Naieme (Naomi) is curious about how technology and spatial data can reveal new perspectives on how cities function. After earning her first Master’s in Urban Planning, she worked as a planner focusing on spatial and policy research. Her work involved translating complex urban issues into practical solutions, which strengthened her appreciation for the human dimensions of planning. To broaden her perspective and explore the analytical side of cities, she returned to university to study Urban Data Science and Analytics at the University of Leeds. Building on her experience and newly developed skills, Naieme dreams of integrating data, spatial insight, and human understanding to help cities evolve toward more sustainable and responsive environments.
Dissertation Title: Through a Machine’s Lens: What Deprivation Looks Like in Leeds
Dissertation Synopsis:
Walking down a street often gives an immediate “sense” of the place, which cannot be easily tracked. These perceptions may also not be universal or easy to measure; however, they can reinforce cycles of disadvantage.
In England, it is the Index of Multiple Deprivation (IMD) that provides the standard, policy-driven measure of inequality. While robust, the IMD is complex, expensive, and updated infrequently. Advances in Machine Learning and the availability of Street View Images offer new possibilities for understanding what streets can tell us about inequality.
This project explored whether new tools such as Machine Learning and Street View Images can offer a practical way of estimating and understanding deprivation. Beyond simply predicting deprivation scores, the aim was to uncover and interpret the visible signs of inequality in the streets of Leeds, as a case study for this project.
The main question was simple:
Can machines help us to “see” deprivation across cities?
It is broken down into the following sub-questions:
• Q1. To what extent is deprivation visible and understandable in the physical environment of Leeds?
• Q2. How accurately can machine learning models detect deprivation from SVI?
• Q3. How does model performance vary across different deprivation domains?
• Q4. Can machines identify visual features that shape how we perceive urban inequality?
Multiple datasets, including nearly 16 K GSV, were cleaned, filtered, and linked to official IMD scores and domains, at the LSOA level. Images were embedded into 512-dimensional feature vectors using a Vision Transformer (ViT-B/16) and screened to remove duplicates, low-quality captures, and indoor scenes.
Several ML models were tested on these embeddings, with Support Vector Classification (SVC) emerging as the best performer. The SVC was then refined through parameter tuning, class recategorising, LSOA-level aggregation, and class balancing to improve both accuracy and interpretability.
Also, to improve transparency, saliency maps are used to show the most important parts of each image that influenced the model’s decision.
First Prize – Undergraduate
Conor Nugent – University of Leeds
Biography: I have recently graduated from the University of Leeds with a First Class degree in BA (Hons) Geography with Quantitative Research Methods (Industrial). In my third year, I completed a year in industry at Adept Insight, a Leeds-based strategic network planning firm led by Chris Duley and Julia Williams. During this year in industry, I supported Adept’s consulting and analysis for clients across a range of industries, including forecourts, homewares, trade merchants, and more.
Following my graduation, I have taken my analysis skills into the finance industry, where I now work as an Audit Associate in the Corporate Assurance team at Buzzacott, the UK’s largest single-office accountancy firm, as I study towards the ACA qualification over the next 3 years. Despite the change of industry, I still have a keen interest in geography, location planning, and sustainability, and continue to follow the industry in my spare time.
Dissertation Title: Stranded: A Methodological Framework for Quantifying the Risk of Loneliness in a Digital World
Dissertation Synopsis:
Rationale and Aim: Loneliness poses several physical and mental health risks to older adults. In the context of an ageing UK population and existing healthcare pressures, new area-based measures of loneliness risk are needed to support appropriate policy interventions and the allocation of health and social care resources to areas at high risk. To date, no such measures account for the potential mitigating impact of digital technologies on loneliness and, conversely, the exacerbating impact of the grey digital divide. This research presents a novel framework for estimating the risk of loneliness and digital exclusion amongst older adults at the small-area level in England and Wales.
Method: The Elderly Loneliness & Digital Exclusion Risk index accounts for several indicators of loneliness risk amongst older adults, which, combined with a measure of digital exclusion, estimates the risk of loneliness and digital exclusion amongst older adults at the small-area level. Intended to be applicable throughout England and Wales, the index is applied to Leeds.
Results: Small areas at the highest risk of loneliness and digital exclusion in Leeds are located in the inner-city areas, suburbs, and satellite towns, whilst those at the lowest risk are in the city centre and rural areas. Digital exclusion is highest in the urban fringes and rural areas.
Discussion: The index can be used by various public and tertiary sector service providers to support policy interventions and the allocation of health, social care, and digital resources to areas at high risk. However, the index may be disproportionately affected by deprivation and the exact impact of the digital dimension on risk estimates is unclear as the novelty of the index limits its evaluation against existing measures.
Second Prize – Undergraduate
Saisruthi Catari Dinesh – University of Greenwich
Biography: I’m Saisruthi Catari Dinesh, a Computer Science graduate from the University of Greenwich, specialising in geospatial mapping and urban planning using data analytics and machine learning. My main interest lies in exploring how data can be used to understand communities and improve decision-making in areas such as retail location planning, sustainability, and public services.During my degree, I developed strong technical skills in Python, SQL, Power BI, and QGIS, which I have used in several analytical and visualisation projects. My final-year dissertation focused on identifying optimal areas in London for new gym development, using methods such as Principal Component Analysis (PCA), clustering, and the Analytic Hierarchy Process (AHP). I also applied Random Forest validation to assess and strengthen the reliability of the results.
This project helped me combine data science with geographic analysis to produce clear, evidence-based insights that can support business and urban development decisions. I’m passionate about applying my technical and analytical skills to real-world challenges and contributing to data-driven solutions that make a positive impact.
Dissertation Title: Urban Analytics for Fitness Infrastructure: Identifying Optimal Gym Locations in London Using Multi-Criteria Decision Analysis and Spatial Intelligence
Dissertation Synopsis:
With the growing emphasis on health-conscious lifestyles and the rise of remote working, the demand for fitness infrastructure in urban areas has become increasingly important. This dissertation focuses on the challenge of identifying suitable locations for new gym developments in London, a city marked by socio-economic diversity and complex spatial dynamics. The overall aim of the project is to develop a spatial decision support system that integrates urban analytics, machine learning and multi-criteria decision analysis to support evidence-based gym location planning.
The motivation for this work stems from the limitations of traditional location selection methods, which often fail to capture the complex interplay between demographic, economic and commercial factors. Many existing approaches rely on narrow indicators, such as income or population density, which do not reflect broader patterns of demand, accessibility or competition. This project addresses this gap by building a structured framework that can account for multiple layers of information and present them in a form that supports practical decision-making.
The system was designed using publicly available socio-economic and commercial data, which were organised and processed into decision criteria. These included income, age distribution, disability prevalence, economic activity, remote working, gym density and retail rent. Analytical techniques such as dimensionality reduction, clustering and multi-criteria ranking were applied to structure this information into clear measures of gym suitability across boroughs.
Outputs were delivered as thematic maps created in QGIS and an interactive dashboard in Power BI. These allow users to explore borough-level suitability scores, compare affordability and competition, and visualise opportunity zones. Stakeholders including business owners, investors and urban planners can use these outputs to identify areas with high potential for gym development, support regeneration strategies and improve accessibility to fitness infrastructure across London.
To find out more about the Society for Location Analysis Student Awards, click on the button below