Exploring methods for mapping seasonal population changes using mobile phone data

This study assesses four methods for distributing mobile phone user counts from cell towers into administrative units in Namibia to map seasonal population changes. Using anonymized call data records (CDRs) from 2011-2013, the accuracy of each method was evaluated against census data at multiple administrative levels. Seasonal and trend decomposition using Loess (STL) and multivariate clustering were used to identify seasonal user variation patterns. Results indicate that the best method depends on the administrative unit level, with tower ranges providing the most accurate results at finer levels. The study highlights the importance of careful method selection in CDR analysis to avoid impacting the observed seasonal patterns and conclusions.

Author 1 Name, Author 2 Name, Author 3 Name, Author 4 Name, Author 5 Name, Author 6 Name