Movatterモバイル変換


[0]ホーム

URL:


Type:Package
Title:Estimating Regularized Multi-state Models Using L1 Penalties
Version:0.99
Date:2015-01-12
Author:Holger Reulen
Maintainer:Holger Reulen <hreulen@uni-goettingen.de>
Description:Structured fusion Lasso penalized estimation of multi-state models with the penalty applied to absolute effects and absolute effect differences (i.e., effects on transition-type specific hazard rates).
License:GPL-2 |GPL-3 [expanded from: GPL (≥ 2)]
Imports:Rcpp (≥ 0.11.3)
LinkingTo:Rcpp
Packaged:2015-01-12 12:43:14 UTC; Holger Reulen
NeedsCompilation:yes
Repository:CRAN
Date/Publication:2015-01-12 14:41:14

Calculation of risksets needed for partial likelihood formulationof multistate models.

Description

This function calculates the risksets needed to calculatethe partial likelihood of a multistate model, and/or it's derivatives.

Usage

buildrisksets(entry, exit, trans, event, trace)

Arguments

entry

vector with entry times.

exit

vector with exit times.

trans

vector with transition types.

event

vector with noncensoring event indicators.

trace

logical triggering printout of status information during the fitting process.

Details

This function calculates risksets.

Value

A list of length 2 with elementsCi andRi, eachvectors of lengthn.

Author(s)

Holger Reulen


First derivative of the locally quadratic approximated penalty.

Description

This function calculates the first derivative of thelocally quadratic approximated penalty.

Usage

dapproxpenalty(psv, beta, constant)

Arguments

psv

penalty structure vector that determines the l-th penalty component when multyplied withbeta.

beta

vector of regression coefficients.

constant

constant that is needed for the locally (in theneighborhood of 0) quadratical approximation of the absolute value function.

Details

This function calculates the first derivative of the locallyquadratic approximated penalty.

Value

The value of the derivative.

Author(s)

Holger Reulen

Examples

## Not run: almatrix(psv, beta, constant)

ddlpl.

Description

Second partial derivative of the log partial likelihoodwith respect to the linear predictor.

Usage

ddlpl(b, X, Ri, Ci)

Arguments

b

vector of regression coefficients.

X

design matrix.

Ri

list of lengthn with vectors as list elements,with the i-th element being the riskset belonging to the i-th spell.

Ci

list of lengthn with vectors as list elements,with the i-th element capturing the indexes of risksets in which spelli is included.

Details

This function calculates the second partial derivative of the log partial likelihood.

Value

A vector with second gradients.

Author(s)

Holger Reulen

Examples

## Not run: ddlpl(b, X, Ri, Ci)

First derivative of the Log Partial Likelihood.

Description

Calculates the first partial derivative of the log partiallikelihood with respect to the linear predictor.

Usage

dlpl(event, b, X, Ri, Ci)

Arguments

event

non-censoring event indicator.

b

vector of regression coefficients

X

design matrix

Ri

list of lengthn with vectors as list elements,with the i-th element being the riskset belonging to the i-th spell.

Ci

list of lengthn with vectors as list elements,with the i-th element capturing the indexes of risksets in which spelli is included.

Details

This function calculates the first derivative of the log partial likelihood of a Coxtype multistate model.

Value

A vector with the values of the partial first derivatives of thelog partial likelihood with respect to the regression effects.

Author(s)

Holger Reulen

Examples

## Not run: dlpl(event, b, X, Ri, Ci)

First derivative of the penalty function.

Description

This function implements the first derivative of the penalty function.

Usage

dpenaltyfunction(psv, beta)

Arguments

psv

penalty structure vector.

beta

estimated regression effects.

Details

This function implements the first derivative of thepenalty function with respect to the penalty.The term 'penalty function' is described in detail on p. 4 in Oelker,Tutz (2013): A General Family of Penalties for Combining DifferingTypes of Penalties in Generalized Structured Models.

Value

Value of the first derivative of the penalty function (note: thisis always 1, since the penalty fucntion p(xi)=xi is just the identity).

Author(s)

Holger Reulen

Examples

## Not run: dpenaltyfunction(psv, beta)

Fisher information matrix of the log partial likelihood of amultistate model.

Description

This function provides a fast implementation for the calculation of the Fisher information matrixneeded for the estimation of fusion lasso penalized multi-state models in a piece-wise exponential framework.

Usage

fishercpp(Xcpp, mucpp)

Arguments

Xcpp

...

mucpp

...

Details

...

Value

...

Author(s)

Holger Reulen

Examples

## Not run: fishercpp(Xcpp, mucpp)

Fisher information matrix of the log partial likelihood of amultistate model.

Description

This function calculates the Fisher information matrixneeded for the estimation of multistate models using the Fisherscoring algorithm.

Usage

fisherinfo(beta, X, risksetlist, event)

Arguments

beta

vector of regression coefficients.

X

design matrix.

risksetlist

list of lengthn with vectors as list elements,with the i-th element being the riskset belonging to the i-th spell.

event

non-censoring event indicator.

Details

This function implements the Fisher scoring matrix (i.e., thesecond partial derivative of the log partial likelihood with respectto the components of the regression effect vectorbeta).

Value

Fisher information matrixinfo.

Author(s)

Holger Reulen

Examples

## Not run: fisherinfo(beta, X, risksetlist, event)

Fisher information matrix of the Poisson log likelihood.

Description

This function calculates the Fisher information matrixneeded for the estimation of multistate models using the Fisherscoring algorithm.

Usage

fisherinfoP(mu, X)

Arguments

mu

mu.

X

design matrix.

Details

This function implements the Fisher scoring matrix (i.e., thesecond partial derivative of the log partial likelihood with respectto the components of the regression effect vectorbeta).

Value

Fisher information matrixinfo.

Author(s)

Holger Reulen

Examples

## Not run: fisherinfo(mu, X)

Log Likelihood for Poisson Regression.

Description

Calculates the log likelihood for poisson regression.

Usage

llP(beta, X, event, offset)

Arguments

beta

vector of regression coefficients.

X

design matrix.

event

non-censoring event indicator.

offset

offset.

Details

This function calculates the Poisson log likelihood.

Value

The values of the Poisson log likelihood.

Author(s)

Holger Reulen

Examples

## Not run: llP(beta, X, event, offset)

Log Partial Likelihood.

Description

Calculates the log partial likelihood.

Usage

lpl(beta, X, risksetlist, event)

Arguments

beta

vector of regression coefficients.

X

design matrix.

risksetlist

list of lengthn with vectors as list elements,with the i-th element being the riskset belonging to the i-th spell.

event

non-censoring event indicator.

Details

This function calculates the log partial likelihood of a Cox-type multistate model.

Value

The values of the spell-specific log partial likelihood contributions.

Author(s)

Holger Reulen

Examples

## Not run: lpl(beta, X, risksetlist, event)

penMSM.

Description

L1 penalized estimation of multistate models.

Usage

penMSM(type = "fused", d, X, PSM1, PSM2, lambda1, lambda2, w, betastart, nu = 0.5, tol = 1e-10, max.iter = 50, trace = TRUE, diagnostics = TRUE, family = "coxph", poissonresponse = NULL, poissonoffset = NULL, constant.approx = 1e-8)

Arguments

type

character defining the type of penalty, eitherfused orlasso.

d

data set with variables (mandatory)entry,exit,trans, andevent.

X

design matrix.

PSM1

penalty structure matrix containing the penalty structure vectorspsv asrows (lasso part).

PSM2

penalty structure matrix containing the penalty structure vectorspsv asrows (fusion part).

lambda1

vector with penalty parameters for the respective penaltycomponents (lasso part).

lambda2

vector with penalty parameters for the respective penaltycomponents (fusion part).

w

vector containing weights for the respective penaltycomponents.

betastart

vector containing starting values for beta.

nu

numeric value denoting the weight, i.e. a value between 0 and1, of the Fisher scoring updates.

tol

relative update tolerance for stopping of the estimation algorithm.

max.iter

number of maximum iterations if tlerance is not reached.

trace

logical triggering printout of status information during the fitting process. .

diagnostics

logical triggering that Fisher matrix, score vector, and approximated penalty matrix are returned with the results.

family

character defining the likelihood to be used.

poissonresponse

response values for poisson likelihood (if used).

poissonoffset

offset values for poisson likelihood (if used).

constant.approx

constant for locally squared approximation of the absolute value penalty function.

Details

This function is the core function of this package.It implements L1 penalized estimation of multistate models, withthe penalty applied to absolute effects and absolute effectdifferences on transition-type specific hazard rates.

Value

A list with elementsB (matrix with estimatedeffects),aic (Akaike Information Criterion),gcv (GCVcriterion),df (degrees of freedom), and (ifdiagnostics are requested)F (Fisher matrix),s (score vector), andA (approximated penalty matrix).

Author(s)

Holger Reulen

Examples

## Not run: penMSMtype = "fused", d, X, PSM1, PSM2, lambda1, lambda2, w, betastart, nu = 0.5, tol = 1e-10, max.iter = 50, trace = TRUE, diagnostics = TRUE, family = "coxph", poissonresponse = NULL, poissonoffset = NULL, constant.approx = 1e-8)## End(Not run)

Penalty matrix for L1 penalized estimation of multistate models.

Description

This builds up a penalty matrix needed for thepenalized estimation of multistate models.

Usage

penaltymatrix(lambda, PSM, beta, w, constant)

Arguments

lambda

vector with penalty parameters for the respective penaltycomponents.

PSM

penalty structure matrix containing the penalty structure vectorspsv as rows.

beta

vector of regression coefficients.

w

vector containing weights for the respective penalty components.

constant

constat that is needed for the locally (in theneighborhood of 0) quadratical approximation of the absolute valuefunction.

Details

This function calculates the penalty matrix needed for thepenalized estimation of multistate models.

Value

A penalty matrixplambda.

Author(s)

Holger Reulen

Examples

## Not run: penaltymatrix(lambda, PSM, beta, w, constant)

plmatrix.

Description

This function establishes the single vectors that set upthe penalty matrix in functionpenaltymatrix.

Usage

plmatrix(psv, beta, constant)

Arguments

psv

index vector that determines the l-th penalty componentwhen multiplied withbeta.

beta

vector of regression coefficients.

constant

constant that is needed for the locally (in theneighborhood of 0) quadratical approximation of the absolute valuefunction.

Details

This function calculates the value of the l-th penaltycomponent, which is a locally (in the neighborhood of 0) quadraticalapproximation of the absolute value of a regression coefficient, orthe difference between two coefficients, respectively.

Value

The objectresult takes the value of the l-th penaltycomponent.

Author(s)

Holger Reulen

Examples

## Not run: plmatrix(psv, beta, constant)

Score vector and Fisher information matrix of the Poisson log likelihood.

Description

This function calculates the score vector and the Fisher information matrix needed for theestimation of multistate models using the Fisher scoring algorithm.

Usage

sF(mu, X, event)

Arguments

mu

mu.

X

design matrix.

event

non-censoring event indicator.

Details

This function implements the score vector and Fisher information matrix.

Value

s and F.

Author(s)

Holger Reulen

Examples

## Not run: sF(mu, X, event)

Score vector of the log partial likelihood of a multistatemodel.

Description

This function calculates the score vector needed for theestimation of multistate models using the Fisher scoring algorithm.

Usage

scorevector(beta, X, risksetlist, event)

Arguments

beta

vector of regression coefficients.

X

design matrix.

risksetlist

list of lengthn with vectors as listelements, with the i-th element being the riskset belonging to thei-th spell.

event

non-censoring event indicator.

Details

This function implements the score vector (i.e., thefirst partial derivative of the log partial likelihood with respectto the components of the regression effect vectorbeta).

Value

Score vectorscorevector.

Author(s)

Holger Reulen

Examples

## Not run: scorevector(beta, X, risksetlist, event)

Score vector of the Poisson log likelihood.

Description

This function calculates the score vector needed for theestimation of multistate models using the Fisher scoring algorithm.

Usage

scorevectorP(mu, X, event)

Arguments

mu

mu.

X

design matrix.

event

non-censoring event indicator.

Details

This function implements the score vector (i.e., thefirst partial derivative of the Poisson log likelihood with respectto the components of the regression effect vectorbeta).

Value

Score vectorscorevector.

Author(s)

Holger Reulen

Examples

## Not run: scorevectorP(beta, X, event)

[8]ページ先頭

©2009-2025 Movatter.jp