Main Content

Nonspherical Models

Model or correct effects of heteroscedasticity and correlation

Nonspherical models are linear regression models with a serially correlated or heteroscedastic innovations process. After fitting a ordinary linear regression model, you can determine whether the model is nonspherical by diagnosing the residuals. To address the effects of nonspherical innovations, you can use the hac function to compute heteroscedastic-and-autocorrelation consistent (HAC) estimates of the regression coefficients, or the fgls function to implement feasible generalized least squares (FGLS).

Classes

arimaCreate univariate autoregressive integrated moving average (ARIMA) model
regARIMACreate regression model with ARIMA time series errors

Functions

autocorrSample autocorrelation
lbqtestLjung-Box Q-test for residual autocorrelation
parcorrSample partial autocorrelation
archtestEngle test for residual heteroscedasticity
hacHeteroscedasticity and autocorrelation consistent covariance estimators
fglsFeasible generalized least squares

Topics