Alternative Computational Methods for Estimation of Logit Model Parameters.

Abstract

In this paper we compare three different computational approaches for maximum likelihood estimation in logit situations: iterative proportional fitting, iteratively reweighted least squares as implemented in GLIM , a variant of Newton's method applied in a somewhat different form for loglinear and logit formulations.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1979
Accession Number
ADA070882

Entities

People

  • G. W. Stewart
  • Michael M. Meyer
  • Stephen E. Fienberg

Organizations

  • University of Minnesota

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Computational Science
  • Computations
  • Computer Science
  • Convergence
  • Maryland
  • Maximum Likelihood Estimation
  • Military Research
  • Minnesota
  • Multivariate Analysis
  • New York
  • Numerical Analysis
  • Statistics
  • United States
  • United States Government
  • Universities

Fields of Study

  • Mathematics

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Regression Analysis.