NONPARAMETRIC CLASSIFICATION ERROR PROBABILITIES FOR TWO DIMENSIONAL EXPONENTIAL POPULATIONS.
Abstract
An approach to the classification problem is one that is dependent of the amount of information assumed to be known about the distributions of the populations. It is assumed in this thesis that nothing is known about the distributions for a two population case. The probability of misclassification of an individual Z is presented in general. The approach is carried further to develop explicit forms of the error probability when the two populations are bivariate exponential. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1966
- Accession Number
- AD0488507
Entities
People
- James C. Rydzewski
Organizations
- Naval Postgraduate School