This function assesses statistical significance of a `wcr`

or `wnet`

fit by referring the cross-validation criterion to a permutation distribution.

1 2 3 4 5 6 7 8 | ```
wcr.perm(y, xfuncs, min.scale = 0, nfeatures, ncomp, method = c("pcr", "pls"),
covt = NULL, nrep = 1, nsplit = 1, nfold = 5, nperm = 20,
perm.method = NULL, family = "gaussian", seed.real = NULL,
seed.perm = NULL, ...)
wnet.perm(y, xfuncs, min.scale = 0, nfeatures = NULL, alpha = 1, lambda,
covt = NULL, nrep = 1, nsplit = 1, nfold = 5, nperm = 20,
perm.method = NULL, family = "gaussian", seed.real = NULL,
seed.perm = NULL, ...)
``` |

`y, xfuncs, min.scale, nfeatures, method, covt, family, nsplit, nfold` |
arguments passed to |

`ncomp` |
number of components; passed to |

`alpha, lambda` |
tuning parameters, passed to |

`nrep` |
number of replicates to average over, when computing the real-data CV criterion. |

`nperm` |
number of permutations. The default is suitable for toy applications only. |

`perm.method` |
either NULL or one of -
`"responses"` : permute the response vector`y` . -
`"y.residuals"` : permute the residuals upon regressing`y ~ covt` . -
`"x.residuals"` : permute the residuals upon regressing`xfuncs ~ covt` .
See Details. |

`seed.real` |
the seed for random data division for real data. If |

`seed.perm` |
the seed for random data division for permuted data. If |

`...` |
other arguments passed to |

Permutation tests of this type are discussed, in a classification setting, by Ojala and Garriga (2010). Permuting the responses (`perm.method="responses"`

) is appropriate when regressing on functions/images only, with no scalar covariates. For linear regression with covariates, it is preferable to first regress on the covariates, and then permute the residuals (`perm.method="y.residuals"`

). For logistic regression this is not feasible; but, following Potter (2005), one can instead permute the residuals from a regression of the functions/images on the covariates (`perm.method="x.residuals"`

). When `perm.method=NULL`

(the default), `"responses"`

is used if `covt`

is `NULL`

, and `"x.residuals"`

otherwise.

`cv` |
CV value for the real data (averaged over |

`cv.perm` |
CV values for the permuted data. |

`pvalue` |
p-value for the permutation test. |

Lan Huo

Ojala, M., and Garriga, G. C. (2010). Permutation tests for studying classifier performance. *Journal of Machine Learning Research*, 11, 1833–1863.

Potter, D. M. (2005). A permutation test for inference in logistic regression with small- and moderate-sized data sets. *Statistics in Medicine*, 24, 693–708.

Reiss, P. T., Huo, L., Zhao, Y., Kelly, C., and Ogden, R. T. (2014). Wavelet-domain regression and predictive inference in psychiatric neuroimaging. Available at http://works.bepress.com/phil_reiss/29/

`wcr`

, `wnet`

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | ```
n = 200; d = 64
# Create true coefficient function
ftrue = matrix(0,d,d)
ftrue[40:46,34:38] = 1
# Generate random functional predictors, and scalar responses
ii = array(rnorm(n*d^2), dim=c(n,d,d))
iimat = ii; dim(iimat) = c(n,d^2)
yy = iimat %*% as.vector(ftrue) + rnorm(n, sd=.3)
## Not run:
obj.wnet.perm <- wnet.perm(yy, xfuncs = ii, min.scale = 4, nfeatures = 200, alpha = 1,
nperm = 10)
obj.wcr.perm <- wcr.perm(yy, xfuncs = ii, min.scale = 4, nfeatures = 20, ncomp = 6,
cv1 = TRUE, method = "pls", nperm = 10)
## End(Not run)
``` |

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.

All documentation is copyright its authors; we didn't write any of that.