Description Usage Arguments Details Value Author(s) See Also
Computes l1-penalized least squares estimates using a coordinate wise descent algorithm of Gauss-Newton quadratic approximations using a matrix free algorithm.
1 2 3 |
f |
a |
gr |
a |
cp |
a |
p |
a |
beta |
a |
lambda |
a |
nlambda |
a |
lambda.min.ratio |
a |
penalty.factor |
a |
rho |
a |
c |
a |
reltol |
a |
trace |
a |
N |
a |
The function computes a matrix of parameter estimates. Each column corresponds to a
value of the penalty parameter in lambda
. The estimates are computed in decreasing order
of the penalty parameters, and for each column the previous is used as a warm start.
The algorithm relies on iterative optimization of the l1-penalized Gauss-Newton quadratic approximation using a standard coordinate wise descent algorithm. An outer backtracking step is added to ensure that the algorithm takes descent steps. This algorithm is matrix free, which means that it does not internally operate with the entire derivative of the mean value map. Instead, it relies on auxiliary functions to compute the loss, the gradient of the loss and inner products of columns in the derivative of the mean value map.
This function relies on three auxiliary functions. The loss function f
, its gradient gr
and a third function, cp
, that computes the cross product of the derivative of the mean value map
one column at a time.
The function returns a list with the vector of lambda values as the first entry and the estimated beta parameters as a matrix in the second entry. Each column in the matrix corresponds to one lambda value.
A list
of length 3. The first element is the sequence lambda
, and the second, beta
, is the
matrix of parameter estimates. Each column in beta
corresponds to an entry in lambda
. The third
element is status
, where 0 means convergence, and 1 means termination due to maximal number of interations
reached (for some lambda).
Niels Richard Hansen Niels.R.Hansen@math.ku.dk
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.