Discrimination procedures, small sample performance.
Eaton, Thomas E.
MetadataShow full item record
The general two population discrimination problem is discussed briefly under various situations,, Discrimination procedures using the linear discriminant function and a nonparametric procedure due to Ju L Hodges and Ee Fix. which classifies a random variable to a population on the basis of assigning it to the population which has the nearest observation to an observed value of the random variable are discussed and compared by computing the probabilities of misclassifieation for both procedures when the two populations are normal with equal covariance matrices e Probabilities of misclassifieation are computed for the nonparametric discriminator and the linear discriminant function for two small sample sizes for the case when the two populations being discriminated are exponential,, In this latter case, both discrimination procedures are shown to give high probabilities of misclassifieation for certain values of the parameters of the distribution being discriminated. Regions are given in terms of the parameters of the two exponential distributions where one of the probabilities of error is greater than 0„5> o A more complete investigation for larger sample sizes is recommended for the linear discriminant function and the nonparametric procedure discussed in this paper for the case when the two populations being discriminated are exponential.
Approved for public release, distribution is unlimited
Showing items related by title, author, creator and subject.
A Comparison of Various Non-Parametric Discriminating Procedures When the Populations Are Bivariate Exponentials Lieberman, Jay Edward (Monterey, California. Naval Postgraduate School, 1969-12);A comparison of the error probabilities for various discriminating rules is performed in the two population cases when nothing is known of the populations other than they are bivariate negative exponential. In most cases, ...
Bernhardt, Stefan (1988-09);The Bootstrap method is a nonparametric statistical technique for estimating the sampling distribution of estimators of unknown parameters. While the asymptotic theory for bootstrap is well established, this thesis ...
Costello, Walter Jeremiah (Monterey, California. Naval Postgraduate School, 1980-03);The linear discrete Kalman filter was analyzed using a frequency-domain approach. Process and measurement noise covariances are shown to be critical design parameters which, together with the assumed prior state and ...