Simulated Data

Description

generates data that can be used for simulations

Usage

1
sim.data.ppls(ntrain,ntest,stnr,p,a=NULL,b=NULL)

Arguments

ntrain

number of training observations

ntest

number of test observations

stnr

signal to noise ratio

p

number of predictor variables

a

vector of length 5 that determines the regression problem to be simulated

b

vector of length 5 that determines the regression problem to be simulated

Details

The matrix of training and test data is drawn from a uniform distribution over [-1,1] for each of the p variables. The response is generated via a nonlinear regression model of the form

Y=∑ _{j=1} ^5 f_j(X_j) + \varepsilon

where f_j(x)=a_j x + sin(6 b_jx). The values of a_j and b_j can be specified via a or b. If no values for a or b is given, they are drawn randomly from [-1,1]. The variance of the noise term is chosen such that the signal-to-noise-ratio equals stnr on the training data.

Value

Xtrain

matrix of size ntrain x p

ytrain

vector of lengt ntrain

Xtest

matrix of size ntest x p

ytest

vector of lengt ntest

sigma

standard deviation of the noise term

a

vector that determines the nonlinear function

b

vector that determines the nonlinear function

Author(s)

Nicole Kraemer

References

N. Kraemer, A.-L. Boulsteix, and G. Tutz (2008). Penalized Partial Least Squares with Applications to B-Spline Transformations and Functional Data. Chemometrics and Intelligent Laboratory Systems, 94, 60 - 69. http://dx.doi.org/10.1016/j.chemolab.2008.06.009

See Also

ppls.splines.cv

Examples

1
dummy<-sim.data.ppls(ntrain=50,ntest=200,p=16,stnr=16)