# Testing the additional predictive value of high-dimensional data

### Description

The function `globalboosttest`

implements a permutation-based testing procedure to globally test the (additional) predictive value of a large set of predictors given that a small set of predictors is already available.

### Usage

1 |

### Arguments

`X` |
A n x p matrix or data frame with observations in rows and variables in columns, whose additional predictive value has to be tested. |

`Y` |
Either a n-vector of type factor (if the prediction outcome is binary), or a numeric vector of length n (if the prediction outcome is numeric and uncensored), or a |

`Z` |
A n x q matrix or data frame with observations in rows and variables in columns, on which we want to condition. Note that q should be smaller than n. If |

`nperm` |
The number of permutations used to derived the p-value. |

`mstop` |
A numeric vector giving the number(s) of boosting steps at which the p-value has to be calculated. |

`mstopAIC` |
If |

`pvalueonly` |
Should the function return only the permutation p-value or also the risk for all numbers of boosting steps and all permutations? |

`plot` |
If |

`...` |
Further arguments to be passed to the |

### Details

See Boulesteix and Hothorn (2009) for details on the methodology.
If `mstopAIC=TRUE`

, the number of boosting steps is chosen from 1 to `max(mstop)`

independently of the specific values
included in the vector `mstop`

.

### Value

A list with the following arguments

`riskreal` |
A numeric vector of length |

`riskperm` |
A |

`mstopAIC` |
The number of boosting steps selected using the AIC-based procedure (if |

`pvalue` |
A numeric vector of length |

### Author(s)

Anne-Laure Boulesteix (http://www.ibe.med.uni-muenchen.de/organisation/mitarbeiter/020_professuren/boulesteix/eng.html),

Torsten Hothorn (http://www.statistik.lmu.de/~hothorn/)

### References

A. L. Boulesteix and Torsten Hothorn (2010). Testing the additional predictive value of high-dimensional data. BMC Bioinformatics 10:78.

### Examples

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | ```
# load globalboosttest library
library(globalboosttest)
# load the simulated data with binary outcome
data(simdatabin)
attach(simdatabin)
# Test with 25 permutations
test<-globalboosttest(X=X,Y=Y,Z=Z,nperm=25,mstop=c(100,500,1000))
# load the simulated data with survival outcome
data(simdatasurv)
attach(simdatasurv)
# Test with 25 permutations
test<-globalboosttest(X=X,Y=Surv(time,status),Z=NULL,nperm=25,mstop=c(100,500,1000),mstopAIC=FALSE)
``` |