constant: Constant Classifier

In locClass: Collection of Local Classification Methods

Description

A classifier that always predicts the class with the highest weighted prior probability.

Usage

  constant(x, ...)

  ## S3 method for class 'formula'
 constant(formula, data,
    weights = rep(1, nrow(data)), ..., subset, na.action)

  ## S3 method for class 'data.frame'
 constant(x, ...)

  ## S3 method for class 'matrix'
 constant(x, grouping,
    weights = rep(1, nrow(x)), ..., subset,
    na.action = na.fail)

  ## Default S3 method:
 constant(x, grouping,
    weights = rep(1, nrow(x)), ...)

Arguments

formula	A formula of the form groups ~ x1 + x2 + ..., that is, the response is the grouping factor and the right hand side specifies the (non-factor) discriminators.
data	A data.frame from which variables specified in formula are to be taken.
x	(Required if no formula is given as principal argument.) A matrix or data.frame or Matrix containing the explanatory variables.
grouping	(Required if no formula is given as principal argument.) A factor specifying the class membership for each observation.
weights	Observation weights to be used in the fitting process, must be larger or equal to zero.
...	Further arguments.
subset	An index vector specifying the cases to be used in the training sample. (NOTE: If given, this argument must be named.)
na.action	A function to specify the action to be taken if NAs are found. The default action is first the na.action setting of options and second na.fail if that is unset. An alternative is na.omit, which leads to rejection of cases with missing values on any required variable. (NOTE: If given, this argument must be named.)

Details

This function is rather a helper function needed to combine mixture models and recursive partitioning with a constant classifier. The weighted prior probabilities are calculated as:

p_g = ∑_{n:y_n=g} w_n/∑_n w_n

Value

An object of class "constant", a list containing the following components:

prior	Weighted class prior probabilities.
counts	The number of observations per class.
lev	The class labels (levels of grouping).
N	The number of observations.
weights	The observation weights used in the fitting process.
predictors	The names of the predictor variables.
call	The (matched) function call.

See Also

predict.constant.

Examples

library(mlbench)
data(PimaIndiansDiabetes)

train <- sample(nrow(PimaIndiansDiabetes), 500)

# weighting observations from classes pos and neg according to their
# frequency in the data set:
ws <- as.numeric(1/table(PimaIndiansDiabetes$diabetes)
    [PimaIndiansDiabetes$diabetes])

fit <- constant(diabetes ~ ., data = PimaIndiansDiabetes, weights = ws,
    subset = train)
pred <- predict(fit, newdata = PimaIndiansDiabetes[-train,])
mean(pred$class != PimaIndiansDiabetes$diabetes[-train])

locClass documentation built on May 2, 2019, 5:21 p.m.