Indexing
All Issues
Submission
Author Guidelines
Reviewers
Editorial Members
Publication Fee
Vol.1 , No. 2, Publication Date: Jul. 7, 2014, Page: 27-30
| [1] | Friday Zinzendoff Okwonu, Department of Mathematics and Computer Science, Delta State University, Abraka, Nigeria. |
The conventional Fisher linear discriminant analysis was proposed to investigate separation between two groups of object. This procedure performs optimally if the data set for the two groups is normally distributed and the variance covariance matrices are homoscedastic. When these assumptions are violated, the Fisher’s technique underperforms. To remedy this deficiency, a supervised learning technique based on the Filter linear classification technique is proposed. The comparative classification performance of these techniques is investigated via Monte Carlo Simulation using data set generated from contaminated normal model. The classification results based on the mean of the optimal probability of correct classification indicate that both procedures are unbiased. Although, the analyses revealed that the Filter technique is robust and admissible over the Fisher’s method.
Keywords
Classification, Mean Probability, Unbiased, Robust, Admissible
Reference
| [01] | Fisher RA. The use of multiple measurements in taxonomic problems. Annals of Eugenics.1936; 7:179 - 188. |
| [02] | Kuhn M, Johnson K. Applied predictive modeling. Springer.2013. |
| [03] | Johnson RA, Wichern DW. Applied multivariate statistical analysis. 6th ed. Pearson Prentice Hall: Upper Saddle River. 2007. |
| [04] | Thiago GM. Reduced- rank discriminant analysis.http://tgmstat.wordpress.com/2013/12/12/reduced-rank-discriminant-analysis. 2013;1-3. |
| [05] | Huberty CJ. Discriminant analysis. Review of Education Research. 1975; 45:543-598. |
| [06] | Filzmoser P, Hron K. Outlier detection for compositional data using robust methods. Mathematical Geosciences. 2008; 40: 233-248. |
| [07] | Basak I. Robust M-estimation in discriminant analysis. The Indian Journal of Statistics. 1998; 60:246-268. |
| [08] | Devlin SJ, Gnanadesikan R, Kettenring JR. Robust estimation of dispersion matrices and principal components. Journal American Statistiscal Association. 1981;76:354-362. |
| [09] | Hubert M, Rousseeuw PJ, Van Aelst S. High breakdown robust multivariate methods. Statistical Science.2008;23; 92-119. |
| [10] | Jin J, An J. Robust discriminant analysis and its application to Identify protein coding regions of rice genes. Mathematical Biosciences. 2011; 232: 96-100. |
| [11] | Kim SJ, Magnani A, Boyd SP. Robust Fisher discriminant analysis. Advances in Neural Information Processing System.2005; 18:659-666. |
| [12] | Pires AM, Branco JA. Generalization of Fisher's linear discriminant. www.math.ist.utl.pt/~apires/PDF/APJB_RP96.pdf, 1996. |
| [13] | Roelant E, Van Aelst S, Williems G. The minimum weighted covariance determinant estimator. Metrika. 2009; 70:177-204. |
| [14] | Wu G, Chen C, Yan X. Modified minimum covariance determinant estimator and its application to outlier detection of chemical process data. Journal of Applied Statistics. 2011; 38:1007-1020. |
| [15] | Linnet K. On the sensitivity of linear discriminant analysis to sampling variation and analytic errors. Computers and Biomedical Research. 1988; 21:158-168. |
| [16] | Zuo Y. Robust location and scatter estimators in multivariate analysis. WSPC/Trim Size:9in x6in for Review.2005; 0-31. |
| [17] | Okwonu FZ, Othman AR. Heteroscedastic variance covariance matrices for unbiased two groups linear classification methods. Applied Mathematical Sciences. 2013; 7: 6855-6865. |
| [18] | Okwonu FZ, Othman AR. Effect of heteroscedastic variance covariance matrices on two groups linear classification techniques. Journal of Mathematics and System Science.2014;4: 133-138. |
| [19] | Okwonu FZ, Dieng H, Othman AR, Hu OS. Classification of aede adult mosquitoes in two distinct groups based on Fisher linear discriminant analysis and FZOARO techniques. Mathematical Theory and Modelling. 2012;2:22-30. |
