Description Usage Arguments Details Value Examples
Implements the equivalent of
pca
.
This function preprocesses the data as specified by the user,
then calls ppcapM or bpcapM, and finally handles this output
to return a list. One element of the output is a pcaRes object.
1 2 3 4 5 6 7 8 9 10 11 12 |
myMat |
|
nPcs |
|
method |
|
seed |
|
threshold |
|
maxIterations |
|
center |
|
scale |
|
loglike |
|
verbose |
|
See ppcapM
and bpcapM
for
the algorithm specifics. loglike
indicates whether
log-likelihood values for the resulting estimates should
be computed. This can be useful to compare different algorithms.
A list
of 5 or 7 elements, depending on the value
of loglike
:
matrix
– the estimated loadings.
numeric
– the estimated isotropic variance.
matrix
– the estimated covariance matrix.
numeric
– the estimated mean vector.
numeric
– the log-likelihood value
of the observed data given the estimated parameters.
numeric
– the log-likelihood value
of the imputed data given the estimated parameters.
class
–
see pcaRes.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 | # simulate a dataset from a zero mean factor model X = Wz + epsilon
# start off by generating a random binary connectivity matrix
n.factors <- 5
n.genes <- 200
# with dense connectivity
# set.seed(20)
conn.mat <- matrix(rbinom(n = n.genes*n.factors,
size = 1, prob = 0.7), c(n.genes, n.factors))
# now generate a loadings matrix from this connectivity
loading.gen <- function(x){
ifelse(x==0, 0, rnorm(1, 0, 1))
}
W <- apply(conn.mat, c(1, 2), loading.gen)
# generate factor matrix
n.samples <- 100
z <- replicate(n.samples, rnorm(n.factors, 0, 1))
# generate a noise matrix
sigma.sq <- 0.1
epsilon <- replicate(n.samples, rnorm(n.genes, 0, sqrt(sigma.sq)))
# by the ppca equations this gives us the data matrix
X <- W%*%z + epsilon
WWt <- tcrossprod(W)
Sigma <- WWt + diag(sigma.sq, n.genes)
# select 10% of entries to make missing values
missFrac <- 0.1
inds <- sample(x = 1:length(X),
size = ceiling(length(X)*missFrac),
replace = FALSE)
# replace them with NAs in the dataset
missing.dataset <- X
missing.dataset[inds] <- NA
# run ppca
ppm <- pcapM(t(missing.dataset), nPcs=5, method="bpca", seed=2009,
maxIterations=1000, center=TRUE, loglike=TRUE, verbose=TRUE)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.