generateInitialUnsupervisedLabels

For a dataset with no observed labels, generate a vector of
initial labels based on the stick-breaking prior for component weights.

Integrative clustering and semi-supervised prediction.
This package includes a 'C++' implementation of both semi-supervised and
unsupervised Multiple Dataset Integration (MDI; Kirk et al., 2012
<https://doi-org.ezp.lib.cam.ac.uk/10.1093/bioinformatics/bts595>), an
extension of Bayesian mixture models to the integrative (multiple dataset or
multi-modal) setting such as multi-omics analysis. The package also includes
Bayesian mixture models. Densities allowed within MDI and the mixture model
are Gaussian with full and diagonal covariance matrices, categorical, Gaussian
with a Gaussian process (GP) prior on the mean parameter. The GP model
implemented canonly handle data with common time points of equal separation.
The Gaussian and GP models can be augmented with a global multivariate t
distribution to handle outliers.

Stephen Coleman

mdir

Bayesian Model-Based Clustering of Multi-Modal Data

Paul Kirk

Chris Wallace

Oliver Crook

generateInitialUnsupervisedLabels function

<dl><dt>N</dt>
<dd>The number of items modelled in the dataset.</dd>
<dt>alpha</dt>
<dd>The concentration parameter.</dd>
<dt>K</dt>
<dd>The number of components modelled.</dd></dl>

Arguments

Generate initial unsupervised labels — generateInitialUnsupervisedLabels

<dl>

<dt>N</dt>
<dd>The number of items modelled in the dataset.</dd>


<dt>alpha</dt>
<dd>The concentration parameter.</dd>


<dt>K</dt>
<dd>The number of components modelled.</dd>

</dl>

generateInitialUnsupervisedLabels: Generate initial unsupervised labels

Description

Usage

Value

Arguments