Abstract
Two kernel-based approaches to discriminant analysis are considered: the traditional one where kernels are used to estimate the distribution of the predictor variables given the group and a direct kernel method where kernels are used to estimate the a posteriori probabilities directly. For both approaches cross-validatory choice of smoothing parameters is based on various loss functions which are directly connected to the separation of groups. Comparison with parametric models shows the improvement gained by the more flexible kernel approaches.
Similar content being viewed by others
References
Aitchison J, Aitken C (1976) Multivariate binary discrimination by the kernel method. Biometrika 63:413–420
Anderson JA (1982) Logistic discrimination. In: Krishnaiah PR, Kanal LN (Eds) Handbook of Statistics, Volume II: Classification, pattern recognition and reduction of dimensionality. North-Holland Amsterdam
Anderson JA, Phillips RR (1981) Regression, discrimination and measurement models for ordered categorical variables. Applied Statistics 30:22–31
Bowman AW (1980) A note on consistency of the kernel method for the analysis of categorical data. Biometrika 67:682–684
Bowman AW, Hall P, Titterington DM (1984) Cross-validation in nonparametric estimation of probabilities and probability densities. Biometrika 71:341–351
Brown PJ, Rundell WK (1985) Kernel estimates for categorical data. Technometrics 27:293–299
Campell MK, Donner A (1989) Classification efficiency of multinomial logistic regression relative to ordinal logistic regression. J Am Stat Ass 84:587–591
Dawid AP (1976) Properties of diagnostic data distributions. Biometrics 32:647–658
Gasser T, Müller H (1984) Nonparametric estimation of regression functions and their devivatives. Scand J Statist 11:171–185
Goldstein M, Dillon WR (1978) Discrete discriminant analysis. New York, John Wiley & Sons
Habbema JDF, Hermans J, Remme J (1978) Variable kernel density estimation in discriminant analysis. In: Corster LCA, Hermans J (eds) Compstat 1978 pp 178–185 Physica Verlag Vienna
Hall P (1981) On nonparametric multivariate binary discrimination. Biometrika 68:287–294
Hall P, Wand M (1988) On nonparametric discrimination using density differences. Biometrika 75:541–547
Hand DJ (1982) Kernel discriminant analysis. Research Studies Press, John Wiley & Sons Ltd
Hand DJ (1982) Statistical pattern recognition of binary variables. In: Kittler J, Fu KS, Pau LF (eds) Pattern recognition theory and applications 19–33. Reidel Publishing Company
Lachenbruch PA, Goldstein M (1979) Discriminant analysis. Biometrics 35:69–85
Lauder IJ (1983) Direct kernel assessment of diagnostic probabilities. Biometrika 70:251–256
McCullagh P (1980) Regression models for ordinal data. JR Statist Soc B 42:109–142
Staniswalis JG (1989) The kernel estimate of a regression function in likelihood based models. J Am Stat Soc 84:276–283
Titterington DM (1980) A comparative study of kernel-based density estimates for categorical data. Technometrics 22:259–268
Titterington DM, Murray GD, Murray LS, Spiegelhalter DJ, Skene AM, Habbema JDF, Gelpke GJ (1981) Comparison of discrimination techniques applied to a complex data set of head injured patients. JR Statist Soc A 144:145–175
Tutz G (1986) An alternative choice of smoothing for kernel-based density estimates in discrete discriminant analysis. Biometrika 73:405–411
Tutz G (1989a) On cross-validation for discrete kernel estimates in discrimination. Commun of Statistics Theory & Methods 11:4145–4162
Tutz G (1989b) Consistency of cross-validatory choice of smoothing parameters for direct kernel estimates. Regensburger Beiträge zur Statistik & Ökonometrie 20
Tutz G (1990) Log-linear parameterizations in discriminant analysis. ZOR — Methods and Models of Operations Research 34:303–319
Van Ness J, Simpson C (1976) On the effects of dimension in discriminant analysis. Technometrics 18:175–187
Van Ness J (1979) On the effects of dimension in discriminant analysis for unequal covariance populations. Technometrics 21:119–127
Wang M-Ch, Van Ryzin J (1981) A class of smooth estimators for discrete distributions. Biometrika 68:301–309
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Tutz, G., Groß, H. Discrete kernels, loss functions and parametric models in discrete discrimination: A comparative study. ZOR - Methods and Models of Operations Research 42, 217–230 (1995). https://doi.org/10.1007/BF01415754
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01415754