Time Series Methods for Frequently-Collected Data

Course Outline

Day 1
Models prior to testing for serial correlation
Regression, Trends, Seasonality review
How to build a good regression model
Residuals analysis
Trend analysis with sin and cos terms

Time Series Fundamentals
Types of Data, Objectives, Graphics
Stationarity
What is serial correlation? How to test for it?
Effects of autocorrelation on hypothesis tests
Effective sample size

Time series analysis
Nonparametric Tests of temporal correlation
Parametric Tests of temporal correlation
Autocorrelation function
ACF when some data are missing
Partial autocorrelation functions

Time series models
Autoregressive models
Moving average models
ARMA models
Model estimation
Diagnostic plots: ACFs and PACFs

Day 2
Forecasting
ARMA models for forecasting

Multivariate Time Series
Basic concepts
Cross Correlation Function
Computation & Significance of the CCF

Estimating a WQ variable from another, frequently measured variable
Problems with using regression for forecasts and record fill-in
Line of Organic Correlation vs classical linear regression
Individual estimates vs. forecasting a collection of data (percentiles)
Comparisons to standards and 1:1 lines

Bootstrap methods for time series