The adoration for cities is widespread across the globe. However, as urbanization escalates in the cities of the global south, concerns regarding unsustainable living have become increasingly prominent. Consequently, there is a pressing need to delve deeper into comprehending the essence of cities and their mechanisms. While understanding cities in terms of their physical configurations and the patterns of human spatial interaction has been a subject of multidisciplinary research over the past few centuries, significant advancements in the field of urban science have emerged in the last three decades. Complexity science and geo-computational models have enabled the study of cities as dynamic entities using a bottom-up approach.

This thesis constructs a conceptual framework encompassing urban form, flow, and human behaviour, which is then applied to the city of Delhi to investigate critical urban phenomena. Specifically, it examines commuting behaviour, the spatial distribution of services, typologies of built-up forms, residential location choice, and built-up expansion. In this endeavour, the study aims to provide insights into pivotal questions within urban science. These include understanding why individuals travel longer distances to their workplaces and the factors that influence their choice of travel mode. Additionally, it investigates the spatial distribution of various services throughout the city for different socio-economic neighbourhoods. The impact of urbanization on unsustainable built-up forms is also explored, along with the relationship between density patterns, and city affordability. Moreover, the study explores how urban planning can be made more efficient by incorporating the decision-making processes of planners into simulation models.

To undertake this research, diverse and novel datasets, including primary and secondary sources, were utilized for the city of Delhi. These encompassed field survey data on commuting behaviour; a spatial database containing population, income, and caste information for all residential locations in Delhi; street map data; and land satellite imageries. The study also employed various machine learning methods and spatial-statistical techniques, such as geographically weighted regression, k-means clustering, SHAP method, agent-based model, and neural network model.

The empirical findings presented in the different chapters of this thesis demonstrate that in Delhi, both urban form and flow are interconnected and influenced by human behaviour. The spatial location of households and neighbourhoods within the city plays a significant role, as does the socioeconomic makeup of these areas, in determining commuting behaviour and the spatial distribution of services. From an urban planning perspective, the city exhibits spatial heterogeneity in neighbourhood design, with the majority of neighbourhoods characterized by unsustainable built-up forms. Consequently, monitoring future built-up expansion should be a priority for Delhi's planners. Using an agent-based and neural network model, this study constructs a prioritised growth model that has the potential to showcase how planning interventions can influence future spatial growth and built-up expansion within the city.

Based on the findings of this study, we recommend that future planning interventions in Delhi consider the enhancement of accessibility for low-income groups alongside environmental sustainability.