An Isotonic Regression Algorithm.
Abstract
A 'pooling sets' type of algorithm is developed and shown to be valid for computing an isotonic regression function for a general quasi-order. The method is direct and intuitive. The algorithm works best when the quasi-order is complex and the objective function is nearly isotonic. An example is worked out in detail. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1981
- Accession Number
- ADA107620
Entities
People
- Richard L. Dykstra