estimation: 2 step procedure: Variable selection and estimation and...
In yymmhaha/PackPaper1: Package related to my first paper

Description Usage Arguments Value Examples

Implements hierarchical variable selection and plain lasso (glmnet) selection. The hierarchy comes from SPAMS paper (to be described). Gives beta for different lags. The second step is also implemented here.

1	estimation(X.shift, Y.single, Y.shift, lagX)

`X.shift`	a matrix. Shifted X matrix
`Y.single`	a vector. Response vector
`Y.shift`	a matrix. Shifted response vector (for second step)
`lagX`	a number. Lags of X to be estimated. In our example it's 2,3 or 5

a list of components

`beta.hat`	Estimated regression coefficients for specified lags, using hierarchical penalty
`beta.lasso`	Estimated regression coefficients for specified lags using lasso
`phi.hat`	Estimated AR coefficients for residual fitting
`beta.refit`	Combination of phi and beta

modelsimple <- estimation(X.shift = DATsimple$Xtr,
Y.single = DATsimple$Ytr.single,
Y.shift = DATsimple$Ytr,
lagX = parameter.of.estimation$lag.to.estX)