orderedLasso (version 1.7.1)

timeLagLasso: Fit a time-lag lasso

Description

Fit a time-lag lasso model. Builds a regression model with multiple predictors, where an ordered constraint is imposed on each predictor.

Usage

timeLagLasso(x, y, lambda, maxlag, intercept = TRUE, standardize = TRUE,
  beta_pos = NULL, beta_neg = NULL, b0 = NULL, strongly.ordered = TRUE,
  maxiter = 500, inneriter = 100, iter.gg = 100, method = c("Solve.QP",
  "GG"), trace = FALSE, epsilon = 1e-05)

Arguments

x

A matrix of predictors, where the rows are the samples and the columns are the predictors.

y

A vector of observations, where length(y) equals nrow(x).

lambda

Regularization parameter(>0).

maxlag

maximum time-lag specified by user.

intercept

TRUE if there is an intercept in the model. center=FALSE should almost never be used. This option is available for special uses only.

standardize

Standardize the data matrix x.

beta_pos

Optional vector of initialization of the positive part of the coefficients.

beta_neg

Optional vector of initialization of the negative part of the coefficients.

b0

Initialization of the intercept.

strongly.ordered

An option which allows users to order each predictor coefficients non-increasing in absolute value. Details can be seen in the orderedLasso description; default TRUE.

maxiter

Maximum iterations run by time-lag lasso. Initialized to 500.

inneriter

Maximum iterations run by orderedLasso. Initialized to 100.

iter.gg

Maximum iterations run by generalized gradient. Intialized to 100

method

Two options available, Solve.QP and Generalized Gradient

trace

Output option; trace = TRUE gives verbose output.

epsilon

Error tolerance parameter for convergence criterion; default 1e-5.

Value

bp

Estimated coefficients of the positive part

bn

Estimated coefficients of the negative part

beta

Estimated coefficients of predictors, which are equal to bp - bn

b0

Estimated intercept if there is one in the model

fitted

Fitted values of y

maxlag

Maximum time-lag

p

The number of predictors

type

Type of model fit, "gaussian" in this case

Details

First transfer the data matrix x to the correct format corresponding to the maxlag specified by user. Then use coordinate descent method by calling orderedLasso to get updates for each predictor.

Examples

# NOT RUN {
set.seed(3)
n = 50
maxlag = 5
num_rows_needed = n + maxlag + 1
sigma = 4
x = matrix(rnorm(num_rows_needed * 4), nrow = num_rows_needed)
x_new = time_lag_matrix(x, maxlag)
b = c(3,1,1,0,0,
      4,1,0,0,0,
      3,2,1,0,0,
      1,0,0,0,0)
y = x_new %*% b + sigma* rnorm(nrow(x_new))
y = as.vector(y)
y = c(y, rnorm(maxlag + 1))
model1 = timeLagLasso(x = x, y = y, lambda =  1, maxlag = maxlag, 
          method = "Solve.QP", strongly.ordered = TRUE)
# }