Study shows over-sampling biases prediction results on imbalanced datasets.
problem Over-optimistic prediction results on imbalanced data.
method Applying over-sampling before partitioning training and testing sets.
result Over-sampling causes biased results and reduces predictive performance.