|
Activity Number:
|
445
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 2, 2016 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
|
Abstract #321075
|
|
|
Title:
|
Predicting Binary Outcome with Unequal Misclassification Cost
|
|
Author(s):
|
Shuchismita Sarkar* and Michael D. Porter
|
|
Companies:
|
University of Alabama and University of Alabama
|
|
Keywords:
|
cost sensitve learning ;
binary classification ;
cost proportionate sampling ;
boosting
|
|
Abstract:
|
While many real-world classification problems involve unequal misclassification costs, most classification methods are based on the assumption of equal penalties for false positives and false negatives. There are two primary approaches for addressing the unequal costs: (i) cost-proportionate sampling and (ii) modification of the classification algorithm to consider a cost matrix. This study compares the relative performance of these two approaches and proposes a combined approach that provides improved performance. The process is illustrated by comparing the performance of the proposed method to existing cost-sensitive classification techniques on a mining data.
|
Authors who are presenting talks have a * after their name.
