wfmm_setup: Function to setup the design matrix
In jpetrovich02/BFOFR: Bayesian Function-on-Function Regression for Multilevel Data

This function performs the specified basis operations on X (e.g. discrete wavelet transformation, PCA). This function should be used prior to using the wfmm executable file, which performs the BFOFR as described in Meyer, et. al. (2015)

1	wfmm_setup(Y, X, Z, MCMCspecs, wavespecsx, pca, testspecs, pcaspecsx)

`Y`	N-by-T matrix of observed response functions. N is the total number of observed curves, T is the number of grid points at which the curves were observed.
`X`	N-by-V matrix of observed predictor functions. N is the total number of observed curves, V is the number of grid points at which the curves were observed.
`Z`	N-by-n random effects design matrix. N is the total number of observed curves, n is the number of subjects.
`MCMCspecs`	A list of specifications for the MCMC algorithm used to estimate the beta coefficients. Needed for the wfmm executable file.
`wavespecsx`	A list of specifications for the wavelet transformation of X. This list must include: "comp.level", a proportion supplied as the compression level, "alpha", "xticks", "nlevels", "wavelet", "wtmode", and "ndim"
`pca`	An indicator of whether or not PCA is to be performed on X after the DWT. 1 indicates yes, 0 no. Default is 1.
`testspecs`	A list of test specifications for posterior inference. This list must contain "twosample", an indicator of whether the design matrix has two samples, and FDR, a vector containing cutoffs for the Bayesian FDR procedure.
`pcaspecsx`	If pca = 1, a list of specifications for the PCA performed on X. The list should contain "pca_level", the percentage of explained variation chosen as a cutoff to determine how many PC's to keep, and "pca_cutoff", a vector of such pca level values to consider. If the argument is not supplied, defaults to "pca_cutoff" = c(0.9, 0.95, 0.975, 0.99, 0.995, 0.999, 0.9999) and "pca_level" = 0.99.

This function returns a list of 3 items: "model", "xspecs", and "testspecs". "model" contains "Z" (unchanged), "X" which is now the basis-space design matrix, and "C". "xspecs" is a list including the wavespecs and pcaspecs used for x. "testspecs" includes the original "testspecs" as well as splits the "FDR" vector into "alf" and "delt".

jpetrovich02/BFOFR documentation built on May 19, 2019, 10:44 p.m.