Handling imbalanced data in multi-class classification problem

Machine Learning Classification
#1

I have multi-class classification problem and data is heavily skewed. My target variable (y) has 3 classes and their % in data is as follows: 0=3% - 1=85% - -1=12%

I am looking for Packages in R which can do multi-class oversampling, under sampling or both the techniques. I tried using ROSE package in R but it works only for binary class problems.

#5

You can try SMOTE.
SMOTE over or under samples the data by generating the observations if needed.Hence, most of the time SMOTE performs better than other sampling techniques.

Here is some sample code on how to do this using SMOTE

from imblearn.over_sampling import SMOTE
sm = SMOTE(random_state=99, ratio = 1.0)
x_train, y_train = sm.fit_sample(X_var, target_class)
print(pandas.value_counts(y_train))
#verify class distribution here

In addition to this you can try using Caret to perform a undersample or oversample of more than two two classes.