esmote

esmote, an R package including fast SMOTE algorithm.

This is part of my undergraduate final year project. Which provide a really fast implementation of SMOTE algorithm.

If you have any concerns please contact me: [email protected]

Some functions are still underconstruction. Ex. I developed a semi-supervised autoencoder to deal with high-dimensional data. However I did not provide a well documented R warpper. You may refer to the source code (./tests/testPer.R).

Installation

First install devtools in R:

install.packages("devtools")

Install package via install_github:

library(devtools)
install_github('HMJiangGatech/ESmote')

Practical Example

This package contains some test data, such as hand written digits data.

newlabel = digitsTrainLabel;
newlabel[newlabel>0] = 1;
newID = sample(60000);
timestart<-Sys.time();

newdata<-esmote::Smote(digitsTrain[newID,],newlabel[newID], algorithm="rp_forest");

timeend<-Sys.time()
runningtime<-timeend-timestart
print(runningtime)

Compared to other packasges such as: DMwR, smotefamily, it is extremely fast.