A statistical method for finding the dependency relationships between a set of variables irregularly sampled in time

Abstract

This paper proposes a method that allows the reconstruction of an equally-spaced time series from another series recorded with irregular times. By using an irregular autoregressive model with parameters varying in time for the unequally spaced time series, new values can be interpolated to find an equispaced series similar to the original series. Then, the model is applied to a set of six irregular time series obtained from an ecological study conducted on the water column of the monitoring station Paraná, located in the Be-agle Channel, Tierra de Fuego (Argentina). The study was designed to know the dependence relationships among the study variables which allow interpreting the temporal dynamics of the system.