time_table_lars: Perform lasso regression on a time.table

In rossklin/dynpan: Dynamics estimation for multiple time series

Description

Perform lasso regression on a time.table

Usage

time_table_lars(x, y = NULL, idxs = NULL, adaptive = 0.5,
  normalise = TRUE, use.auxiliary = FALSE, input.cols = NULL,
  output.cols = NULL, ..., modelfun = function(x) polySane(x, raw = TRUE,
  degree = 2), has.no.na = FALSE, adaptive.lambda = 0.1)

Arguments

x	time.table that contains predictors and, optionally, dependent variable(s).
y	time.table containing dependent variable(s).
idxs	index/time values to include, defaults to all complete cases
adaptive	exponent used for the adaptive weights set to NULL or 0 to disable (defaults to 0.5)
use.auxiliary	whether to include auxiliary values
input.cols	column(s) of x to use for computing covariate(s)
output.cols	column(s) of x or y to use as dependent varaiable(s)
...	additional arguments to pass to modelfun
modelfun	function that produces the actual covariates used in the linear regression
has.no.na	whether user guarantees x/y contian no NA values
adaptive.lambda	ridge regression shrinkage parameter to use when calculating adaptive lasso weights

rossklin/dynpan documentation built on May 27, 2019, 11:39 p.m.