Piecewise Geometric Estimation of a Survival Function.

Abstract

This document describes a statistical procedure that uses incomplete data to estimate failure rate and survival functions. Although the procedure is designed for discrete distributions, it applied in the continuous case also. This description is expository and therefore contains no proofs: they are provided by Mimmack (1985). The procedure is based on the assumption of a piecewise constant failure rate. The resultant survival function estimator is a piecewise geometric function, denoted the Piecewise Geometric Estimator (PEGE). The PEGE is the discrete version of the piecewise exponential estimators proposed independently by Kitchin, Langberg and Proschan (1983) and Whittemore and Keller (1983), and it is a generalization of an estimator of Umholtz (1984) who considers complete data taken from an exponential distribution. The PEGE is consistent and asymptotically normal under conditions more general than those of the model of random censorship. Although the PEGE and the widely used Kaplan-Meier estimator (KME) are asumptotically equivalent and generally interlace, the PEGE is expected to perform better than the KME in terms of small sample properties. The PEGE is attractive to users because it is computationally simple and realistic in that it decreases at every possible failure time; it therefore no only has the appearance of a survival function, but also provides a realistic estimate of the failure rate function. The KME, in contrast, is a step function. Keywords: pilot studies; Monte Carlo method.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1985
Accession Number
ADA161322

Entities

People

  • Frank Proschan
  • Gillian M. Mimmack

Organizations

  • Florida State University

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Biomedical Research
  • Consistency
  • Data Science
  • Discrete Distribution
  • Distribution Functions
  • Estimators
  • Information Science
  • Pilot Studies
  • Probability
  • Probability Distributions
  • Random Variables
  • Scientific Research
  • Simulations
  • Statistics
  • Step Functions
  • Universities

Fields of Study

  • Mathematics

Readers

  • Statistical inference.