Smote is an abbreviation for Synthetic Minority Oversampling TEchnique. This tag refers to the oversampling method used commonly in machine learning to balance the class distributions in datasets by introducing new minority class examples.
In machine learning, most classifiers works assuming that the classes given in the training set are roughly balanced. When classes are imbalanced, classifiers tend towards predicting the majority class.
One way to overcome this is to carry out an interpolation among neighboring minority class instances and generate artificial samples.
Useful references:
One of the earlier publications on SMOTE: chawla et al 2002
One review on SMOTE: Fernández et al 2017
Influence of datasets on SMOTTE: Skryjomski et al 2017
Python toolbox for imbalanced datasets: Lemaˆıtre et al 2017
