Error-dependency relationships for the naïve Bayes classifier with binary features

Ludmila I Kuncheva; Zoë S J Hoare

doi:10.1109/TPAMI.2007.70845

Error-dependency relationships for the naïve Bayes classifier with binary features

IEEE Trans Pattern Anal Mach Intell. 2008 Apr;30(4):735-40. doi: 10.1109/TPAMI.2007.70845.

Authors

Ludmila I Kuncheva¹, Zoë S J Hoare

Affiliation

¹ School of Computer Science, Bangor University, Sean Street, Bangor, Gwynedd, UK. [email protected]

PMID: 18276977
DOI: 10.1109/TPAMI.2007.70845

Abstract

We derive a tight dependency-related bound on the difference between the Naïve Bayes (NB) error and Bayes error for two binary features and two equiprobable classes. A measure of discrepancy of feature dependencies is proposed for multiple features. Its correlation with NB is shown using 23 real data sets.

MeSH terms

Algorithms*
Artificial Intelligence*
Bayes Theorem*
Computer Simulation
Data Interpretation, Statistical*
Logistic Models
Pattern Recognition, Automated / methods*
Reproducibility of Results
Sensitivity and Specificity
Signal Processing, Computer-Assisted*