lfmm_ridge_CV: Cross validation of LFMM estimates with ridge penalty

In bcm-uga/MatrixFactorizationR: Latent Factor Mixed Models

Description

This function splits the data set into a train set and a test set, and returns a prediction error. The function lfmm_ridge is run with the train set and the prediction error is evaluated from the test set.

Usage

lfmm_ridge_CV(Y, X, n.fold.row, n.fold.col, lambdas, Ks)

Arguments

Y	a response variable matrix with n rows and p columns. Each column corresponds to a distinct response variable (e.g., SNP genotype, gene expression level, beta-normalized methylation profile, etc). Response variables must be encoded as numeric.
X	an explanatory variable matrix with n rows and d columns. Each column corresponds to a distinct explanatory variable (eg. phenotype). Explanatory variables must be encoded as numeric.
n.fold.row	number of cross-validation folds along rows.
lambdas	a list of numeric values for the regularization parameter.
Ks	a list of integer for the number of latent factors in the regression model.
p.fold.col	number of cross-validation folds along columns.

Details

The response variable matrix Y and the explanatory variables X are centered.

Value

a dataframe containing prediction errors for all values of lambda and K

Author(s)

cayek, francoio

See Also

lfmm_ridge

Examples

library(ggplot2)
library(lfmm)

 ## sample data
 K <- 3
 dat <- lfmm_sampler(n = 100, p = 1000, K = K,
                     outlier.prop = 0.1,
                     cs = c(0.8),
                     sigma = 0.2,
                     B.sd = 1.0,
                     U.sd = 1.0,
                     V.sd = 1.0)

 ## run cross validation
 errs <- lfmm_ridge_CV(Y = dat$Y,
                         X = dat$X,
                         n.fold.row = 5,
                         n.fold.col = 5,
                         lambdas = c(1e-10, 1, 1e20),
                         Ks = c(1,2,3,4,5,6))

 ## plot error
 ggplot(errs, aes(y = err, x = as.factor(K))) +
   geom_boxplot() +
   facet_grid(lambda ~ ., scale = "free")

 ggplot(errs, aes(y = err, x = as.factor(lambda))) +
   geom_boxplot() +
   facet_grid(K ~ ., scales = "free")

bcm-uga/MatrixFactorizationR documentation built on June 17, 2020, 6:55 p.m.