Description Usage Arguments Examples
View source: R/cobra_validate.r
cobra_validate
performs an MM algorithm wrapper to do parameter selection.
1 2 3 |
X |
Data matrix |
E_row |
Edge-incidence matrix for row graph |
E_col |
Edge-incidence matrix for column graph |
w_row |
Vector of weights for row graph |
w_col |
Vector of weights for column graph |
gamma |
Regularization parameter for shrinkage |
Lambda_row |
Initial guess of row Langrage multipliers |
Lambda_col |
Initial guess of column Langrage multipliers |
fraction |
Fraction of entries for hold out |
max_iter |
Maximum number of iterations |
tol |
Stopping criterion |
max_iter_inner |
Maximum number of inner cobra iterations |
tol_inner |
Stopping criterion for inner cobra loop |
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 | ## Create bicluster path
## Example: Lung
X <- lung
X <- X - mean(X)
X <- X/norm(X,'f')
## Create annotation for heatmap
types <- colnames(lung)
ty <- as.numeric(factor(types))
cols <- rainbow(4)
YlGnBu5 <- c('#ffffd9','#c7e9b4','#41b6c4','#225ea8','#081d58')
hmcols <- colorRampPalette(YlGnBu5)(256)
## Construct weights and edge-incidence matrices
phi <- 0.5; k <- 5
wts <- gkn_weights(X,phi=phi,k_row=k,k_col=k)
w_row <- wts$w_row
w_col <- wts$w_col
E_row <- wts$E_row
E_col <- wts$E_col
## Connected Components of Row and Column Graphs
wts$nRowComp
wts$nColComp
#### Initialize path parameters and structures
nGamma <- 7
gammaSeq <- 10**seq(0,1,length.out=nGamma)
## Generate solution path
sol <- cobra_validate(X,E_row,E_col,w_row,w_col,gammaSeq,fraction=0.01)
## Plot validation error
verr <- sol$validation_error
plot(verr)
## Heatmap of data smoothed at the model selected to minimize validation error
ix <- which.min(verr)
groups_row <- sol$groups_row[[ix]]
groups_col <- sol$groups_col[[ix]]
M <- biclust_smooth(X,groups_row,groups_col)
heatmap(M,col=hmcols,labRow=NA,labCol=NA,ColSideCol=cols[ty])
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.