Forecasting with Univariate Box - Jenkins Models: Concepts and CasesExplains the concepts and use of univariate Box-Jenkins/ARIMA analysis and forecasting through 15 case studies. Cases show how to build good ARIMA models in a step-by-step manner using real data. Also includes examples of model misspecification. Provides guidance to alternative models and discusses reasons for choosing one over another. |
Common terms and phrases
ABS Z ERR absolute t-values acf and pacf acf in Figure ADJUSTED RMSE ALAN PANKRATZ Copyright analysis AR(l ARIMA model autocorrelation at lag autocovariances available data BACKCASTS Box and Jenkins Box-Jenkins calculated chi-squared statistic COEF T-VAL LAG COEFFICIENT ESTIMATE STD constant term correlated cutoff to zero DATA COUNT data in Figure DATA SERIES diagnostic checking diagnostic-checking results different from zero DUE TO PRESENCE equation ESTIMATE STD ERROR estimated acf estimated autocorrelation estimated coefficients Estimation and diagnostic-checking example expected value exponential decay FIND SSR Forecasting With Univariate identification stage LOST DUE MA(l MEAN ABS ordered pairs pacf’s parameters partial autocorrelation practical warning level random shocks random variables residual acf residual autocorrelations results for model seasonal differencing seasonal lags seasonal pattern spike at lag standard error stationarity STD DEV STD ERROR T-VALUE theoretical acf’s variance VERTICAL AXIS INTERVAL