Novel robust time series analysis for long-term and short-term prediction

Abstract Nonlinear phenomena are universal in ecology. However, their inference and prediction generally difficult because of autocorrelation outliers. A traditional least squares method for parameter estimation is capable improving short-term by estimating autocorrelation, whereas it has weakness to outliers consequently worse long-term prediction. In contrast, a robust regression approach, such as the absolute deviations method, alleviates influence potentially better prediction, makes accurately possibly leads We propose new approach that estimates reduces then compare with conventional methods using simulated data real ecological data. Simulations analysis demonstrate ability nonlinear problems spawner–recruitment The provides nearly unbiased even highly contaminated extreme outliers, other fail estimate accurately.