mnemk: Learn the number of components K and optimize the mixture.

Description Usage Arguments Value Author(s) Examples

View source: R/mnems.r

Description

High level function for learning the number of components k, if unknown.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
mnemk(
  D,
  ks = seq_len(5),
  man = FALSE,
  degree = 4,
  logtype = 2,
  pen = 2,
  useF = FALSE,
  Fnorm = FALSE,
  ...
)

Arguments

D

data with cells indexing the columns and features (E-genes) indexing the rows

ks

vector of number of components k to test

man

logical. manual data penalty, e.g. man=TRUE and pen=2 for an approximation of the Akaike Information Criterion

degree

different degree of penalty for complexity: positive entries of transitively reduced phis or phi^r (degree=0), phi^r and mixture components minus one k-1 (1), phi^r, k-1 and positive entries of thetas (2), positive entries of transitively closed phis or phi^t, k-1 (3), phi^t, theta, k-1 (4, default), all entries of phis, thetas and k-1 (5)

logtype

logarithm type of the data (e.g. 2 for log2 data or exp(1) for natural)

pen

penalty weight for the data (e.g. pen=2 for approximate Akaike Information Criterion)

useF

use F (see publication) as complexity instead of phi and theta

Fnorm

normalize complexity of F, i.e. if two components have the same entry in F, it is only counted once

...

additional parameters for the mnem main function

Value

list containing the result of the best k as an mnem object and the raw and penalized log likelihoods

Author(s)

Martin Pirkl

Examples

1
2
3
4
sim <- simData(Sgenes = 3, Egenes = 2, Nems = 2, mw = c(0.4,0.6))
data <- (sim$data - 0.5)/0.5
data <- data + rnorm(length(data), 0, 1)
result <- mnemk(data, ks = seq_len(2), starts = 1)

mnem documentation built on Nov. 18, 2020, 2 a.m.