Modelling Locally Changing Variance Structured Time Series Data By Using Breakpoints Bootstrap Filtering
Rajan Lamichhane · ODU Digital Commons (Old Dominion University) · 2019
Stochastic processes have applications in many areas such as oceanography and engineering. Special classes of such processes deal with time series of sparse data. Studies in such cases focus in the analysis, construction and prediction in parametric models. Here, we assume several non-linear time series with additive noise components, and the model fitting is proposed in two stages. The first stage identifies the density using all the clusters information, without specifying any prior knowledge of the underlying distribution function of the time series. The effect of covariates is controlled by fitting the linear regression model with serially correlated errors. In the second stage, we partition the time series into consecutive non-overlapping intervals of quasi stationary increments where the coefficients shift from one stable regression relationship to a different one using a breakpoints detection algorithm. These breakpoints are estimated by minimizing the likelihood from the residuals. We approach time series prediction through the mixture distribution of combined error components. Parameter estimation of mixture distribution is done by using the EM algorithm. We apply the method to fish otolith data influenced by various environmental conditions and get estimation of parameters for the model.