Flowminder – Geospatial Analysis of Call Data Records

As part of Flowminder’s humanitarian work in a conflict zone in the middle east, I extracted descriptive statistics and created geospatial visualizations from de-identified call data records provided by an operator of a mobile phone network.

This project posed some great challenges around infrastructure. The core data source, falling within the terabyte range, needed to be prepared for large-scale statistical analysis in a computationally efficient manner and integrated with a heterogeneous set of supplemental data sources for contextualization.

As part of my exploratory analysis, I extracted descriptive statistics and created geospatial visualizations of various characteristics of the data.

The project was conducted through a multidisciplinary team involving data scientists as well as subject matter experts. There was a strong culture of contextualizing results through subject matter expertise to get to actionable insight.

In addition to my role within the project team on this project, I was Flowminder’s point person to focus on machine learning aspects of the work going on throughout the organisation.