Effective Statistical Tests for Detection Models
Abstract
A Detection Model is an entity which calculates an instantaneous probability of detection of a target by a hunter, from the values of variables which describe the environment and the actions of both hunter and target, including past history, if appropriate. Given such a model, and a succession of non-identical trials which terminate at detection or after a given period of time (whichever occurs first), it is desired to test the adequacy of the model. An approach to this problem is presented, based upon recognizing the set of trials as a non-homogeneous Poisson process. Ways to improve the power of such tests by rearranging various segments of the trials are presented and discussed, including proper implementation of the tests using a digital computer. Extensions to the problem of improving the model and/or devising a new model are briefly discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 17, 1972
- Accession Number
- AD0740776
Entities
People
- James A. Lechner
Organizations
- George Washington University