MLSeqModelInfo-class: 'MLSeqModelInfo' object

In dncR/MLSeq: Machine Learning Interface for RNA-Seq Data

Description

For classification, this is the subclass for the MLSeq class. This object contains all the information about classification model.

Details

Objects can be created by calls of the form MLSeqModelInfo(...). This type of objects is created as a result of classify function of MLSeq package. It is then used in predictClassify function for predicting the class labels of new samples.

Slots

method, transformation, normalization:

these slots store the classification method, transformation technique and normalization method respectively. See notes for details.

preProcessing:

See classify for details.

ref:

a character string indicating the reference category for cases (diseased subject, tumor sample, etc.)

control:

a list with controlling parameters for classification task.

confusionMat:

confusion table and accuracy measures for the predictions.

trainedModel:

an object of MLSeq.train class. It contains the trained model. See notes for details.

trainParameters:

a list with training parameters from final model. These parameters are used for test set before predicting class labels.

call:

a call object for classification task.

Note

method, transformation, normalization slots give the information on classifier, transformation and normalization techniques. Since all possible pairs of transformation and normalization are not available in practice, we specify appropriate transformations and normalization techniques with preProcessing argument in classify function. Finally, the information on normalization and transformation is extracted from preProcessing argument.

MLSeq.train is a union class of train from caret package, voom.train and discrete.train from MLSeq package. See related class manuals for details.

See Also

train, voom.train-class, discrete.train-class

dncR/MLSeq documentation built on May 17, 2020, 6:45 p.m.