generate_imbalanced_data: Generate an imbalanced data set.

In RomeroBarata/bimba: Sampling Algorithms for Two-Class Imbalanced Data Sets

Description

generate_imbalanced_data is a simple function to generate a two-class imbalanced data set.

Usage

generate_imbalanced_data(num_examples = 100L, num_features = 2L,
  imbalance_ratio = 5, noise_maj = 0.05, noise_min = 0.1, seed = NULL)

Arguments

num_examples	Total number of examples in the data set.
num_features	Total number of features in the data set.
imbalance_ratio	Ratio of the number of examples in the majority class to the number of examples in the minority class.
noise_maj	Fraction of the minority class that is mislabelled as majority class.
noise_min	Fraction of the majority class that is mislabelled as minority class.
seed	Integer value for reproducibility purposes.

Details

The imbalanced data set generated has two classes where the majority class comes from a multivariate normal distribution with mean zero and unitary standard deviation for all features and the minority class comes from a multivariate normal distribution with mean two and unitary standard deviation for all features.

The total number of examples and the dimensionality of the data are chosen through the num_examples and num_features arguments. The imbalance_ratio argument together with num_examples determines the exact number of examples in the majority and minority classes. To simulate noise in the data, approximately noise_min examples in the majority class are labelled as minority class examples and approximately noise_maj examples in the minority class are labelled as majority class examples. noise_maj and noise_min are fractions.

Value

A data frame containing an imbalanced two-class data set.

RomeroBarata/bimba documentation built on May 17, 2019, 8:03 a.m.