SMOTE stands for Synthetic Minority Over-sampling Technique. SMOTE creates new synthetic points in order to have an equal balance of the classes. This is another alternative for solving the "class imbalance problems".
Understanding SMOTE:
Achieving Balanced Classes : Using the distances between the closest neighbors of the minority class, SMOTE creates synthetic points in between these distances in order to reach an equal balance between the minority and majority class.
Effect: More information is retained since we don't have to delete any rows unlike in random undersampling(where we remove some data samples of majority class to achieve class balance).
Accuracy - Time Tradeoff: More the number of data samples, more the training time which tends to increase performance. Less the number of data samples, less the information and thus less probable of decent performance.
Loading comments...