Description Usage Arguments Value Examples
The input matrices should have samples as the rows and features as columns. A kernel will computed across all samples in the first matrix with respect to the samples in the second matrix. The two matrices must have the same features. If all features are binary 0,1, then the Jaccard similarity kernel will be used, otherwise, a Gaussian kernel with standard deviation equal to s times the mean euclidean distances between samples in the second matrix. If there are samples with all NA values, they will not appear in the kernel matrix columns. The row for that sample will just be all NAs.
1 |
m1 |
matrix on samples X features to compute kernels on |
m2 |
matrix of samples X features to compute kernels with respect to. |
s |
numeric multiplier of standard deviation for the Gaussian kernels (default:1). |
matrix of similarities between rows of m1 and rows of m2.
1 2 3 4 5 6 7 8 9 10 11 | m1 <- matrix(rnorm(200), 8, 25, dimnames=list(paste0('sample.', 1:8), paste0('feat.', 1:25)))
m2 <- matrix(rnorm(100), 4, 25, dimnames=list(paste0('sample.', 9:12), paste0('feat.', 1:25)))
kernelize(m1, m1)
kernelize(m1, m1, s=.5)
kernelize(m2, m1)
m1 <- matrix(rbinom(200, 1, .5), 8, 25,
dimnames=list(paste0('sample.', 1:8), paste0('feat.', 1:25)))
m2 <- matrix(rbinom(25, 1, .5), 1, 25,
dimnames=list(c('sample.9'), paste0('feat.', 1:25)))
kernelize(m1, m1)
kernelize(m2, m1)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.