Important step forward to recurrence analysis of data with uncertainties

In a new publication we propose a novel approach for analysing data with uncertainties, which is typical, e.g., in palaeoclimate observations. A time series of single values is replaced by a time series of probabilities; the binary matrix representation of recurrences is then replaced by  a matrix of probability values, which represent that a recurrence at a certain time appears. This approach is the base for a novel transition test, applying concepts from recurrence and complex network analysis. We demonstrate the potentials on examples from present-day climate (ENSO), palaeoclimate (Indian summer monsoon), and stock markets.

probability series

Time series of probability distributions of proxy values, allowing for uncertainty analysis.

Open access: