Multiple Change Point Estimation in Stationary Ergodic Time Series
This work addresses the challenge of detecting distributional changes in time series for applications in fields like finance or signal processing, offering a general and nonparametric approach, though it appears incremental as it builds on existing change point estimation methods.
The paper tackles the problem of estimating multiple change points in time series data generated by unknown stationary ergodic distributions without modeling or independence assumptions, proposing a novel nonparametric method that is shown to be asymptotically consistent and computationally efficient.
Given a heterogeneous time-series sample, the objective is to find points in time (called change points) where the probability distribution generating the data has changed. The data are assumed to have been generated by arbitrary unknown stationary ergodic distributions. No modelling, independence or mixing assumptions are made. A novel, computationally efficient, nonparametric method is proposed, and is shown to be asymptotically consistent in this general framework. The theoretical results are complemented with experimental evaluations.