Overly Optimistic Prediction Results on Imbalanced Data - Flaws and Benefits of Over-sampling

Explore the intricacies of over-sampling techniques in machine learning through this 22-minute Launchpad video. Delve into the paper "Overly Optimistic Prediction Results on Imbalanced Data: Flaws and Benefits of Over-sampling" to understand the potential pitfalls and advantages of using over-sampling methods on imbalanced datasets. Examine case studies, including the TPEHGDB and APTOS datasets, to gain insights into the effects of over-sampling on model performance. Learn about the comparison of results across different studies and the implications for real-world applications. Gain valuable knowledge on how to critically evaluate prediction results and avoid common mistakes when dealing with imbalanced data in machine learning projects.

Intro
ABSTRACT
NOTES ON DATA - TPEHGDB
NOTES ON STUDIES AND FLAWS
NOTES ON STUDIES REVIEWED
NOTES ON OVERSAMPUNG
PAPER COMPARISON OF RESULTS
NOTES ON DATA - APTOS
MY EXPERIMENTS - MODEL RESULTS
FURTHER READING