What to Optimize?

  • Ezra Hauer


Parameters estimated by weighted least squares produced bad fits. The question is whether the fit can be improved by maximizing likelihood or using some alternative objective function. The concepts of likelihood, likelihood function, and maximum likelihood estimation will be explained and illustrated. In preparation for model development some commonly used likelihood functions will be given and implemented on a C-F spreadsheet. When the purpose of SPF development is to support practical road safety management, quality of fit rather than maximization of likelihood may be the preferred objective. The use of such alternative objective functions will prove both simple and attractive.


Model Equation Likelihood Function Road Segment Negative Binomial Distribution Negative Binomial 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. Aitken AC (1935) On least squares and linear combinations of observations. Proc Roy Soc Edinburgh 55:42–48CrossRefzbMATHGoogle Scholar
  2. Cameron AC, Trivedi PK (1998) Regression analysis of count data. Cambridge University Press, CambridgeCrossRefzbMATHGoogle Scholar
  3. Charnes A, Frome EWL, Yu PL (1976) The equivalence of generalized least squares and maximum likelihood estimates in the exponential family. J Am Stat Assoc 71(353):169–171MathSciNetCrossRefzbMATHGoogle Scholar
  4. Edwards AWF (1972) Likelihood. Cambridge University Press, CambridgezbMATHGoogle Scholar
  5. Edwards AWF (1974) The history of likelihood. Int Stat Rev 42(1):9–15MathSciNetCrossRefzbMATHGoogle Scholar
  6. Geyer (2003) Stat 5102 notes: Fisher information and confidence intervals using maximum likelihood. . Downloaded 4/13/2014
  7. Guo G (1996) Negative multinomial regression models for clustered event counts. Sociol Methodol 26:113–132CrossRefGoogle Scholar
  8. Hauer E (2001) Overdispersion in modeling crashes on road sections and in empirical Bayes estimation. Accid Anal Prev 33:799–808CrossRefGoogle Scholar
  9. Hyndman RJ, Koehler AB (2006) Another look at measures of forecast accuracy. Int J Forecast 22:679–688CrossRefGoogle Scholar
  10. Schermelleh-Engel K, Moosbrugger H, Müller H (2003) Evaluating the fit of structural equation models: tests of significance and descriptive goodness-of-fit measures. Methods Psychol Res Online 8(2):23–74Google Scholar
  11. Stigler SM (1986) The history of statistics. Harvard University Press, CambridgezbMATHGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  • Ezra Hauer
    • 1
  1. 1.University of TorontoTorontoCanada

Personalised recommendations