New fast algorithms for error rate-based stepwise variable selection in discriminant analysis

SIAM Journal on Scientific Computing

Volumn 22, Issue 3, 2001, Pages 1036-1052

  Aeberhard, S ^a   De Vel, O Y ^a   Coomans, D H ^a  

^a JAMES COOK UNIVERSITY   (Australia)

Author keywords

Classification; Classifier based variable selection; Dimensionality reduction

Indexed keywords

COMPUTATIONAL COMPLEXITY; COMPUTATIONAL METHODS; ERROR ANALYSIS; HEURISTIC METHODS; LINEARIZATION; MULTIVARIABLE SYSTEMS; SET THEORY; VARIATIONAL TECHNIQUES;

CLASSIFIER BASED VARIABLE SELECTION; DISCRIMINANT ANALYSIS; LINEAR CLASSIFIERS; MAHALANOBIS DISTANCE; MULTIVARIATE PREDICTIVE ANALYSIS; QUADRATIC CLASSIFIERS; STEPWISE VARIABLE SELECTION; WILK DISTANCE;

ALGORITHMS;

EID: 0034973218     PISSN: 10648275     EISSN: None     Source Type: Journal    
DOI: 10.1137/S1064827596300784     Document Type: Article

References (17)

1. New fast algorithms for variable selection based on classifier performance
2. Introduction to Statistical Pattern Recognition
3. Discriminant analysis and statistical pattern recognition
4. Regularized discriminant analysis
5. On selecting variables and assessing their performance in linear discriminant analysis
6. Matrix computations
7. Selection of variables in discriminant analysis by F-statistic and error rate
8. Discrimination and classification
9. Error-rate estimation in multiple-group linear discriminant analysis
10. Applied discriminant analysis
11. Feature set search algorithms
12. Estimation of error rates in discriminant analysis
13. Variable selection techniques in discriminant analysis: I. Description, and II. Allocation
14. UCI repository of machine learning databases [machine-readable data repository]
15. A cautionary note on selection of variables in discriminant analysis
16. Small sample size effects in statistical pattern recognition: Recommendations for practitioners
17. An evaluation of smoothed classification error-rate estimators