Clairvoyant: AdaBoost with Cost-Enabled Cost-Sensitive Classifier for Customer Churn Prediction

Table 2

Definitions of symbols of the equations.

H	Final hypothesis/model combining all weak hypotheses
	The hypothesis/model at iterations
	The prediction of the data point by the hypothesis/model
	The probability distribution of the data point
	The new probability of the data point at iteration
	Hypothesis’s weight for gross misclassification error at iteration
	Hypothesis’s weight for high-risk (false-negative) error at iteration
	Cost of misclassification for false-negative error specified in the input cost matrix
	, the normalization