Nonparametric strategies applied to time series analysis
Claudio Pizzi, Francesca Parpinel · ARCA (Università Ca' Foscari Venezia) · 2006
In this work we present a nonparametric test to detect nonlinearity in time series. The test is based on permutation methods and essentially it is a distributional comparison between two sub-populations. We evaluate the significance nominal level and the power of tests by simulation considering differen t linear and nonlinear models. As benchmarking we use some well known nonlinearity tests.