Data Science & Big Data Analytics

The company has extensive experience in advanced spatial analysis and data visualisation using Geographic Information Systems (GIS). Our work in this area integrates spatial and temporal high-resolution big data with other data sources which are more static to deliver a comprehensive understanding of urban dynamics and mobility behaviours. MIC-HUB develops urban activation maps, origin–destination flow maps, and temporal signatures that describe inflow and outflow patterns around major urban attractors.

Internationally recognised metrics, such as PTALs (Public Transport Accessibility Levels) developed by Transport for London, as well as Space Syntax indicators like betweenness are incorporated in the spatial analyses to assess network connectivity and efficiency. MIC-HUB also uses evaluation metrics like Walkscore or other proprietary metrics to assess walkability of the urban space, with the aim of developing active mobility and enhancing the quality of urban environment.

MIC-HUB develops machine learning (ML) algorithms and models to analyse urban mobility phenomena and their interaction with the built environment, providing predictive tools to support decision-making and sustainable mobility planning. MIC-HUB’s models can predict traffic flows, travel behaviours, and road safety risk levels by processing big data from sensors, GPS, social media, and traffic simulations. This contributes to greater efficiency, safety and sustainability in transport systems. ML algorithms help decode the relationships between mobility and the built environment in terms of urban density, land use, and street design, revealing how these factors influence travel choices, distances, transport modes, and externalities such as crash rates and emission levels.

MIC-HUB uses supervised learning algorithms, including neural networks, to develop predictive models that assess traffic volumes, network speeds, and accident rates, based on the urban and sociodemographic characteristics of each study area. Unsupervised machine learning techniques use cluster analysis and dimensionality reduction to identify unknown patterns. These analyses help to reveal hidden structures within complex, heterogeneous data on mobility and other urban characteristics, such as chains of human movement.