Regression contains linear models with independently and identically distributed errors and for errors with heteroscedasticity or autocorrelation
The statistical model is assumed to be
, where
depending on the assumption on , we have currently four classes available
All regression models define the same methods and follow the same structure, and can be used in a similar fashion. Some of them contain additional model spedific methods and attributes.
GLS is the superclass of the other regression classes.
Class hierachy: TODO
yule_walker is not a full model class, but a function that estimate the parameters of a univariate autoregressive process, AR(p). It is used in GLSAR, but it can also be used independently of any models. yule_walker only calculates the estimates and the standard deviation of the lag parameters but not the additional regression statistics. We hope to include yule-walker in future in a separate univariate time series class. A similar result can be obtained with GLSAR if only the constant is included as regressors. In this case the parameter estimates of the lag estimates are not reported, however additional statistics, for example aic, become available.
The following is more verbose description of the attributes which is mostly common to all regression classes
General reference for regression models:
Analysis.” 2nd. Ed., Wiley, 1992.
Econometrics references for regression models:
see also the examples and the tests folders
OLS(endog[, exog]) | A simple ordinary least squares model. |
GLS(endog, exog[, sigma]) | Generalized least squares model with a general covariance structure. |
WLS(endog, exog[, weights]) | A regression model with diagonal but non-identity covariance structure. |
GLSAR(endog[, exog, rho]) | A regression model with an AR(p) covariance structure. |
yule_walker(X[, order, method, df, inv]) | Estimate AR(p) parameters from a sequence X using Yule-Walker equation. |