bc_estimate: Wrapper for Binary Classification Estimation

In rettopnivek/binclass: Personalized functions for binary classification.

Description

Fits different types of models for binary classification to a training subset of data, and then computes fit metrics over a test subset.

Usage

bc_estimate(type, dat, control = NULL)

is.bc_estimate(x)

## S3 method for class 'bc_estimate'
coef(x, int = FALSE, sig = T)

## S3 method for class 'bc_estimate'
subset(x, train = F, metric = "AUC")

## S3 method for class 'bc_estimate'
levels(x)

## S3 method for class 'bc_estimate'
dimnames(x)

## S3 method for class 'bc_estimate'
print(x, digits = 2, metric = "AUC")

## S3 method for class 'bc_estimate'
features(x)

Arguments

type	The type of fitting algorithm. Options include glm and glmnet.
dat	An R object of class train_test.
control	An optional list of additional estimation parameters.

Details

The control argument is a list that can supply any of the following components:

• col_sel = An index for the subset of predictors to use;

• second_pass = Logical; if TRUE, refits a glm model with only the significant predictors;

• prev_fit = Allows previous output from glm or glmnet to be passed in;

• error_rate = The cutoff for determining whether a variable is statistically significant or not;

• prev_coef = An optional vector with previous coefficients to use as offsets in a null model for glm;

• alpha = The mixing proportion ridge (0) versus lasso (1) regression;

• nfolds = The number of folds for cross-validation step to select the best penalty term for glmnet;

• costf = The type of cost function to use for the cv.glmnet function, where options include 'auc' (Area under the curve), 'mae' (mean absolute error), 'class' (misclassification error), and 'deviance'.

The method subset can extract the specified fit metric for either the training (train = TRUE) or test subsets. The features method extracts the labels for the predictors deemed significant.

Value

An R object of class 'bc_estimate'.

Examples

# Simulate data
sim = bc_simulate( 300, 8, 4 )
# Create training and test data
index = cv_index( 3, 300 )
dat = train_test( 3, index, sim$y, sim$X )
fit = bc_estimate( 'glm', dat )
fit_2 = bc_estimate( 'glmnet', dat )

rettopnivek/binclass documentation built on May 13, 2019, 4:46 p.m.