The method for solving two types of errors in customer segmentation on unbalanced data

Authors:
Zou Peng;Hao Yuanyuan
Affiliations:
Harbin Institute of Technology, P. R. China;Harbin Institute of Technology, P. R. China
Venue:
Proceedings of the 10th international conference on Electronic commerce
Year:
2008

Citing 11
Cited 0

Probability, random processes, and estimation theory for engineers

Probability, random processes, and estimation theory for engineers
Generalized predictive control—Part I. The basic algorithm

Automatica (Journal of IFAC)
Solving the quadratic programming problem arising in support vector classification

Advances in kernel methods
Cost-Sensitive Learning by Cost-Proportionate Example Weighting

ICDM '03 Proceedings of the Third IEEE International Conference on Data Mining
Decision trees with minimal costs

ICML '04 Proceedings of the twenty-first international conference on Machine learning
Integrating AHP and data mining for product recommendation based on customer lifetime value

Information and Management
A threshold varying bisection method for cost sensitive learning in neural networks

Expert Systems with Applications: An International Journal
Cost-sensitive classification: empirical evaluation of a hybrid genetic decision tree induction algorithm

Journal of Artificial Intelligence Research
The foundations of cost-sensitive learning

IJCAI'01 Proceedings of the 17th international joint conference on Artificial intelligence - Volume 2
Support vector machine techniques for nonlinear equalization

IEEE Transactions on Signal Processing
Support vector machines and the multiple hypothesis test problem

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Two types of errors of "rejecting true and accepting false" are inevitable in customer value segmentation. The traditional data mining method which is on the total accuracy rate can not reflect the influence caused by the great difference of misclassification costs and unbalanced quantity distribution of customers who have various values. The thesis proposes the cost-sensitive SVM (support vector machine) classifier by presenting misclassification cost function based on customer value, which is evaluated by the function of the exceptional lost. The data test result proves that the method can control the different types of errors distribution with various cost of misclassification accurately, reduce the total misclassification cost, and distinguish the customer value effectively.