PlatypusML_classification: Core ML for crossvalidation

In Platypus: Single-Cell Immune Repertoire and Gene Expression Analysis

Core ML for crossvalidation

Description

This PlatypusML_classification function takes as input encoded features obtained using the PlatypusML_extract_features function. The function runs cross validation on a specified number of folds for different classification models and reports the AUC scores and ROC curves.

Usage

PlatypusML_classification(features, cv.folds, balancing, proportion)

Arguments

features	Matrix. Output of the PlatypusML_extract_features function, containing the desired label in the last column.
cv.folds	Integer. The number of folds to be used in cross validation
balancing	Boolean. Whether to perform class balancing. Defaults to TRUE.
proportion	Vector of size 2 (floats between 0 and 1 that need to sum up to 1). Specifies the proportions for the two classes. The smaller proportion will be assigned to the minority class by default. Defaults to c(0.5,0.5).

Value

This function returns a list containing [["combined"]] summary plot with ROC & confusion matrices, [["ROC"]] the ROC curve, [["confusion"]] confusion matrices for each classifier.

Examples

## Not run: 
To classify and obtain the performance of different models,
using extracted and encoded features.

#extract features
features_VDJ_GP33_binder <- PlatypusML_feature_extraction_VDJ(VGM = VGM,
which.features = c("VDJ_cdr3s_nt"),
which.encoding = c("kmer"),
parameters.encoding.nt = c(3),
which.label = "GP33_binder")

#classify
classifier_GP33_binder <- classification(features = features_VDJ_GP33_binder,
cv.folds = 5,
balancing = TRUE)

#view summary
classifier_GP33_binder$combined

## End(Not run)

Platypus documentation built on Aug. 15, 2022, 9:08 a.m.