Imbalanced Learning
lastmod 2021-08-01created 2021-08-011 edits
Imbalanced Learning overview
In cases such as abnormal transaction detection, the share of transaction records judged to fall outside the normal range will be far smaller than that of normal transactions. This data imbalance becomes an important issue when trying to find such abnormal transactions with ordinary algorithms, and this post discusses imbalanced learning.
Apart from methods that resolve class imbalance within the algorithm itself, Over-Sampling and Under-Sampling are the most representative methods.