Radius-SMOTE (RadiusSMOTE, 2021)

Resampling with RadiusSMOTE.

Usage

RadiusSMOTE(x, y, k = 5, ovRate = 1, ...)

Arguments

x

feature matrix or data.frame.

y

a factor class variable. Can have more than two classes.

k

number of neighbors used to decide whether a sample is safe (a sample is safe when more than k / 2 of its k nearest neighbors in the whole data are from its own class). Default is 5.

ovRate

Oversampling rate. Class \(j\) receives round(ovRate * (n_max - n_j)) synthetic samples, where n_max is the size of the largest class (halves are rounded up). Default is 1 meaning full balance. A vector with one value per class is also accepted.

...

not used.

Value

a list with resampled dataset.

x_new

Resampled feature matrix.

y_new

Resampled target variable.

x_syn

Generated synthetic feature data.

y_syn

Generated synthetic label data.

Details

Year: 2021. Samples of the class that are safe are used as centers. Synthetic samples are generated on the line between a center and its nearest sample of the other classes, either towards it or in the opposite direction, at a random distance of at most the distance to that sample (the safe radius), so they do not reach the other classes. If no sample is safe, all samples are used with a warning.

Can work with classes more than 2: every class is oversampled against the union of the other classes (one-versus-rest).

References

Pradipta, G. A., Wardoyo, R., Musdholifah, A., & Sanjaya, I. N. H. (2021). Radius-SMOTE: a new oversampling technique of minority samples based on radius distance for learning from imbalanced data. IEEE Access, 9, 74763-74777.

Author

Fatih Saglam, saglamf89@gmail.com

Examples

set.seed(1)
x <- rbind(matrix(rnorm(2000, 3, 1), ncol = 2, nrow = 1000),
           matrix(rnorm(100, 5, 1), ncol = 2, nrow = 50))
y <- as.factor(c(rep("negative", 1000), rep("positive", 50)))

m <- RadiusSMOTE(x = x, y = y, k = 5)
table(m$y_new)
#> 
#> negative positive 
#>     1000     1000 
plot(m$x_new, col = m$y_new)