Mapping the Invisible Boundaries
A Data-Driven Approach to City Delineation
Authors
MSc. Andia VLLAMASI, Department of Information Technology, POLIS University, Tirana, Albania,
Prof. Dr. Tamara LUARASI, Department of Information Technology, POLIS University, Tirana, Albania,
Dr. Luca LEZZERINI, Department of Information Technology, POLIS University, Tirana, Albania,
Abstract
According to estimates, 67% of the world's population is expected to live in urban and sub-urban
areas by 2040, primarily due to ongoing migration from rural areas. This pattern is also very
noticeable in Albania, where a highly populated metropolis that frequently stretches beyond its
administrative borders is the result of rapid urbanisation.
For a long time, researchers and policymakers have struggled to define urban areas. Some of the
traditional methods rely on administrative borders, which often fail to capture the actual economic
and spatial dynamics of cities. Others depend on urban morphology, missing the population
behaviours and needs. To better understand and manage the urban dynamics, this research aims
to try a different method for calculating Tirana's borders. This method will be based on population
distribution, utilising a density-based clustering technique in conjunction with a digital
representation of the urban form. In comparison to the original administrative boundaries, can a
digital, data-driven, density-based algorithm provide a more functionally correct and policyrelevant delineation of metropolitan areas in a mid-sized city like Tirana? And how can we encode
the urban morphology in a digitalised representation that can be both fed into an algorithm and
understood by urban planners?
This project aims to develop a machine learning-based approach that clusters buildings into urban
zones defined by metrics such as density, urban morphology, and geographic distribution. This
approach will lead to the identification of a group of strongly interconnected urban clusters that
better represent the physical environment and distribution of economic activity in Tirana. These
groups will reflect the real functional extent of the city, taking into account its urban form, and
excluding low-density, outlying zones. Additionally, we believe that the vertical land indicator will
provide fresh perspectives on Tirana's urban polycentricity and compactness, which will influence
the design of spatial policies and infrastructure development.
These newly drawn lines will serve as a foundation for further study, enabling more accurate plans
for sustainable development, more focused urban planning, quick detection and reaction to change,
as well as novel opportunities for economic analysis and policymaking.
Keywords
Rapid urbanisation, clustering, urban, machine learning, density
