Analyzing spatial mobility patterns with time-varying graphical lasso: Application to COVID-19 spread.

Abstract

This work applies the time‐varying graphical lasso (TVGL) method, an extension of the traditional graphical lasso approach, to address learning time‐varying graphs from spatiotemporal measurements. Given georeferenced data, the TVGL method can estimate a time‐varying network where an edge represents a partial correlation between two nodes. To achieve this, we use a COVID‐19 data set from the Argentine province of Chaco. As an application, we use the estimated network to study the impact of COVID‐19 confinement measures and evaluate whether the measures produced the expected result.