expres: Naive Classifier of Gene Expression Profiles
In sdnet: Soft-Discretization-Based Bayesian Network Inference

Description Usage Arguments Details Value Author(s)

Implements a naive classifier using soft discretization.

1
2
3

sdnLearn(data, cls, clslevs = NULL, ncats = 3, nodeCats = NULL, quant="uniform", std=TRUE)
sdnPredict(model, data, std=TRUE) 
sdnEvaluate(train, test, ncats = 3, nodeCats = NULL, std=FALSE)

`data`	a numerical `matrix` in row-genes format
`train`	a numerical `matrix` in row-genes format
`test`	a numerical `matrix` in row-genes format
`cls`	a `factor` or `integer`, the sample labels
`clslevs`	an optional vector of labels, should include training data's labels
`ncats`	an `integer`, the number of categories per node
`nodeCats`	a `list`, custom node categories
`quant`	quantization method
`std`	a `logical`, should the data rows be standardized
`model`	a `list` of components for the training model

The model contains a vector of gene names geneset, a vector of sample labels clslevs, class catNetworks: nets, a list of node categories nodeCats and a training quantization model quant.

sdnPredict returns the log-ratio of the class conditional probabilities for each test observation. sdnEvaluate handles 2-class problems and returns the prediction accuracy and predicted classes.

N. Balov

sdnet documentation built on May 2, 2019, 12:43 a.m.