HyperAI

Threshold-moving

Threshold MovementIt refers to adjusting the threshold for classifying categories according to actual conditions, and is often used to solve the problem of category imbalance.

For classification problems, instead of directly predicting the class label of the instance, we can predict the probability value and then specify a threshold to divide the instantiation into positive and negative classes. The threshold is usually specified as 0.5, but the threshold can be moved according to the actual situation, thereby increasing the weight of a certain class to achieve the purpose of solving class imbalance.