THRESHOLDS OF URBAN FORM RESILIENCE INDICATORS INFORMED BY MACHINE LEARNING: AN EXPLORATORY RESEARCH ON HEATING ENERGY DEMAND

Date

2025-8-29

Author

Kutlay, Ecem

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

61
views

0
downloads

Cite This

Two categories of urban resilience are discussed in literature: general and specified. General resilience is defined as the system's overall capacity to survive disturbances, specified resilience is explained as the specific focus to be prepared for. While scholars rarely claim to need both, there's a lack of a quantifiable definition. This thesis aims to integrate the two by considering urban form in a general resilience perspective and defining a specific theme, i.e., the heating energy performance of urban blocks. It explores the extents of morphological indicators for a specific theme, conducting a detailed analysis via machine learning to identify complex relations and performance thresholds. This study initially measures urban form indicators widely recognized in the literature as enhancers of general resilience, using Helsinki Metropolitan Area as a case area. It then critically evaluates the performance ranges of these indicators. To establish a relation between general resilience indicators and energy performance, building-level heating energy (deterministic-archetype) data is aggregated at the urban block level. To model the complex, non-linear relationships among the indicators, the XGBoost ML algorithm outperformed others. Model interpretability is ensured through SHAP (SHapley Additive exPlanations) analysis. Globally, plot granularity, plot number in block and floor space index emerged as the most influential indicators, while local analysis showed variations in their impact and ranking, though largely aligns with global findings. The thesis essentially identifies the various performance ranges and reveals the complex and non-linear/linear relationships between the normatively defined urban form general resilience indicators in relation to heating energy demand.

Subject Keywords

Resilient urban form, Energy performance, General and specified resilience, Machine Learning, Complexity, Non-linearity

URI

https://hdl.handle.net/11511/116209

Collections

Graduate School of Natural and Applied Sciences, Thesis