Time Series Forecasting

Time series forecasting is the use of a model to predict future values based on previously observed values. While regression analysis is often employed in such a way as to test theories that the current values of one or more independent time series affect the current value of another time series, this type of analysis of time series is not called "time series analysis", which focuses on comparing values of a single time series or multiple dependent time series at different points in time.

There are currently no projects for this area of research.

Bayesian Time Series Modelling with Long-Range Dependence

We present a class of models for trend plus stationary component time series, in which the spectral densities of stationary components are represented via non-parametric smoothness priors combined with long-range dependence components. We discuss model fitting and computational issues underlying Bayesian inference under such models, and provide illustration in studies of a climatological time series. These models are of interest to address the questions of existence and extent of apparent long-range effects in time series arising in specific scientific applications.

Time Series Analysis of Diurnal Cycles in Small-Scale Turbulence

Two new time series techniques are employed to study the diurnal effect between thirty to sixty meters depth in the upper equatorial ocean. A modified spectral approach is proposed to identify the period of an unequally spaced series. Based on the exceedances of mixing activities, a new criterion is introduced to identify local nights and days. These two techniques together provide a simple way to demonstrate the existence of the diurnal cycle in deep stratified layers in the ocean. Instead of entertaining complicated models, the proposed approach suggests a simply way to identify the periodicity of an unequally spaced data set of this nature. Similar diurnal effects are detected in the analyses of a subsequent data set.