SMOTE: Synthetic Minority Oversampling TEchnique

In smotefamily: A Collection of Oversampling Techniques for Class Imbalance Problem Based on SMOTE

Synthetic Minority Oversampling TEchnique

Description

Generate synthetic positive instances using SMOTE algorithm

Usage

SMOTE(X, target, K = 5, dup_size = 0)

Arguments

X	A data frame or matrix of numeric-attributed dataset
target	A vector of a target class attribute corresponding to a dataset X.
K	The number of nearest neighbors during sampling process
dup_size	The number or vector representing the desired times of synthetic minority instances over the original number of majority instances

Value

data	A resulting dataset consists of original minority instances, synthetic minority instances and original majority instances with a vector of their respective target class appended at the last column
syn_data	A set of synthetic minority instances with a vector of minority target class appended at the last column
orig_N	A set of original instances whose class is not oversampled with a vector of their target class appended at the last column
orig_P	A set of original instances whose class is oversampled with a vector of their target class appended at the last column
K	The value of parameter K for nearest neighbor process used for generating data
K_all	Unavailable for this method
dup_size	The maximum times of synthetic minority instances over original majority instances in the oversampling
outcast	Unavailable for this method
eps	Unavailable for this method
method	The name of oversampling method used for this generated dataset (SMOTE)

Author(s)

Wacharasak Siriseriwan <[e-mail]>

References

Chawla, N., Bowyer, K., Hall, L. and Kegelmeyer, W. 2002. SMOTE: Synthetic minority oversampling technique. Journal of Artificial Intelligence Research. 16, 321-357.

Examples

    data_example = sample_generator(10000,ratio = 0.80)
	genData = SMOTE(data_example[,-3],data_example[,3])
	genData_2 = SMOTE(data_example[,-3],data_example[,3],K=7)

smotefamily documentation built on May 29, 2024, 7:54 a.m.