simLong {BoostMLR}

Simulate longitudinal data

Description

Simulates longitudinal data from multivariate and univariate longitudinal response model.

Usage

simLong(n = 100,
        ntest = 0,
        N = 5,
        rho = 0.8,
        model = c(1, 2),
        phi = 1,
        q_x = 0,
        q_y = 0,
        type = c("corCompSym", "corAR1", "corSymm", "iid"))

Arguments

n

Requested training sample size.

ntest

Requested test sample size.

N

Parameter controlling number of time points per subject.

rho

Correlation parameter.

model

Requested simulation model.

phi

Variance of measurement error.

q_x

Number of noise covariates.

q_y

Number of noise responses.

type

Type of correlation matrix.

Details

Simulates longitudinal data from multivariate and univariate longitudinal response model. We consider following 2 models:

1. model=1: Simpler linear model consist of three longitudinal responses, y1, y2, and y3 and four covariates x1, x2, x3, and x4. Response y1 is associated with x1 and x4. Response y2 is associated with x2 and x4. Response y3 is associated with x3 and x4.

2. model=2: Relatively complex model consist of single longitudinal response and four covariates. Model includes non-linear relationship between response and covariates and covariate-time interaction.

Value

An invisible list with the following components:

dtaL

List containing the simulated data in the following order: features, time, id and y.

dta

Simulated data given as a data frame.

trn

Index of id values identifying the training data.

Author(s)

Amol Pande and Hemant Ishwaran

References

Pande A., Li L., Rajeswaran J., Ehrlinger J., Kogalur U.B., Blackstone E.H., Ishwaran H. (2017). Boosted multivariate trees for longitudinal data, Machine Learning, 106(2): 277–305.