What you’ll learn
Identify Box-Jenkins autoregressive integrated moving average model integration order through level and differentiated time series first order trend stationary deterministic test and Phillips-Perron unit root test.
Recognize autoregressive integrated moving average model autoregressive and moving average orders through autocorrelation and partial autocorrelation functions.
Estimate autoregressive integrated moving average models such as random walk with drift and differentiated first order autoregressive.
Identify seasonal autoregressive integrated moving average model seasonal integration order through level and seasonally differentiated time series first order seasonal stationary deterministic test.
Estimate seasonal autoregressive integrated moving average models such as seasonal random walk with drift and seasonally differentiated first order autoregressive.
Select non-seasonal or seasonal autoregressive integrated moving average model with lowest Akaike, corrected Akaike and Schwarz Bayesian information loss criteria.
Evaluate autoregressive integrated moving average models forecasting accuracy through mean absolute error, root mean squared error scale-dependent and mean absolute percentage error, mean absolute scaled error scale-independent metrics.
Identify generalized autoregressive conditional heteroscedasticity modelling need through autoregressive integrated moving average model squared residuals or forecasting errors second order stationary Ljung-Box lagged autocorrelation test.
Recognize non-Gaussian generalized autoregressive conditional heteroscedasticity modelling need through autoregressive integrated moving average and generalized autoregressive conditional heteroscedasticity model with highest forecasting accuracy standardized residuals or forecasting errors multiple order stationary Jarque-Bera normality test.
Estimate autoregressive integrated moving average models with residuals or forecasting errors assumed as Gaussian or Student™s t distributed and with Bollerslev simple or Glosten-Jagannathan-Runkle threshold generalized autoregressive conditional heteroscedasticity effects such as random walk with drift and differentiated first order autoregressive.
Assess autoregressive integrated moving average model with highest forecasting accuracy standardized residuals or forecasting errors strong white noise modelling requirement.
“Course Overview
Course Description
Course Overview
Advanced Forecasting Models
Advanced Forecasting Models Data
Course File
Course Overview Slides
“Auto Regressive Integrated Moving Average Models
ARIMA Models Slides
ARIMA Models Overview
First Order Trend Stationary Time Series
ARIMA Models Specification
Random Walk with Drift ARIMA Model
Differentiated First Order Autoregressive ARIMA Model
First Order Seasonal Stationary Time Series
SARIMA Models Specification
Seasonal Random Walk with Drift SARIMA Model
Seasonally Differentiated First Order Autoregressive SARIMA Model
ARIMA Model Selection
ARIMA Models Forecasting Accuracy
“Generalized Auto Regressive Conditional Heteroscedasticity Models
GARCH Models Slides
GARCH Models Overview
Second Order Stationary Time Series
GARCH Models Specification
ARIMA-GARCH Models Estimation
Random Walk with Drift ARIMA-GARCH Model
Differentiated First Order Autoregressive ARIMA-GARCH Model
ARIMA-GJR-GARCH Models Estimation
Random Walk with Drift ARIMA-GJR-GARCH Model
Differentiated First Order Autoregressive ARIMA-GJR-GARCH Model
ARIMA-GARCH Model Selection
ARIMA-GARCH Models Forecasting Accuracy
“Non-Gaussian Generalized Auto Regressive Conditional Heteroscedasticity Models
Non-Gaussian GARCH Models Slides
Non-Gaussian GARCH Models Overview
Multiple Order Stationary Time Series
Non-Gaussian GARCH Models Specification
ARIMA-GARCH-t Models Estimation
Random Walk with Drift ARIMA-GARCH-t Model
Differentiated First Order Autoregressive ARIMA-GARCH-t Model
ARIMA-GJR-GARCH-t Models Estimation
Random Walk with Drift ARIMA-GJR-GARCH-t Model
Differentiated First Order Autoregressive ARIMA-GJR-GARCH-t Model
Non-Gaussian ARIMA-GARCH Model Selection
Non-Gaussian ARIMA-GARCH Models Forecasting Accuracy
Residuals White Noise