s-kganz/scutr

Balancing Multiclass Datasets for Classification Tasks

scutr's Issues

options in usamp_opts/osamp_opts do not work

The options for SCUT do not work.
Error in dist(data[, -which(names(data) == cls_col)], ...) :
unused argument (k = 7)

I run and that is the error.
library(scutr)
data(imbalance)
imbalance <- imbalance[imbalance$class %in% c(2, 3, 19, 20), ]
imbalance$class <- as.numeric(imbalance$class)

plot(imbalance$V1, imbalance$V2, col=imbalance$class)

scutted <- SCUT(imbalance, "class", undersample = undersample_kmeans,
usamp_opts = list(k=7))

It also does not work in a specific data set i am working on. I have problems with the options in general.
Could you explain me if there is a problem in the function or something wrong from my side?

smote did not work for small number of cases

Here is the number of cases in each class (total 22 classes, 460 samples, average is 21)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3 4 2 44 27 10 7 15 15 10 5 8 115 8 3 46 38 6 4 58
21 22
3 29
I ran scut usng default.
command: scutted <- SCUT(zz, "class")
I get the following error:
Error in get.knnx(data, query, k, algorithm) : ANN: ERROR------->
Calls: SCUT ... -> SMOTE -> knearest -> -> get.knnx
In addition: Warning message:
In get.knnx(data, query, k, algorithm) : k should be less than sample size!

It seems for SMOTE to work, K cannot be greater than or equal to the sample size. (in my case it's 2). But to generate 20 samples, k have to >2. is there a way to get around this? Thanks!