Mobilkit: A Python Toolkit for Urban Resilience and Disaster Risk Management Analytics using High Frequency Human Mobility Data

TL;DR

The paper addresses the need for scalable, privacy-conscious analytics on high-frequency mobility data to support disaster risk management. It introduces Mobilkit, an open-source Python toolkit built on Dask that provides an end-to-end framework for DRM analytics on raw GPS pings, from data loading and filtering to displacement analysis and POI metrics. Core contributions include home/work localization via mean-shift, land-use inference via hierarchical clustering, and rapid displacement profiling enabling analyses on large user sets (e.g., 130,000 users over 5 weeks) on commodity hardware, all under an MIT license with accessible documentation. This toolkit offers a practical, replicable path for urban resilience planning and paves the way for richer data modalities and privacy-aware analytics in DRM contexts.

Abstract

Increasingly available high-frequency location datasets derived from smartphones provide unprecedented insight into trajectories of human mobility. These datasets can play a significant and growing role in informing preparedness and response to natural disasters. However, limited tools exist to enable rapid analytics using mobility data, and tend not to be tailored specifically for disaster risk management. We present an open-source, Python-based toolkit designed to conduct replicable and scalable post-disaster analytics using GPS location data. Privacy, system capabilities, and potential expansions of \textit{Mobilkit} are discussed.

Figures (1)

• Figure 1: Analytical capabilities of Mobilkit