Improving Population Estimation with Neural Network Models
Intercensal and postcensal population estimates are essential in federal, state, and local governments planning and resource allocation. Traditionally, linear regression based models are widely used for projecting population distributions in a given region. We constructed population projection models with various types of artificial neural networks. Using historical census data, we tested the performance of the neural network models against the ratio correlation regression model that we have used for the last 20 years. The results indicate that properly trained neural networks outperform the regression model in both model fitting and projection. Among the different neural network models we tested, the fuzzy logic based neural network performed the best.
KeywordsNeural Network Model Hide Node Feedforward Neural Network Mean Absolute Percentage Error State Total
Unable to display preview. Download preview PDF.
- 1.Bishop, C.M.: Neural Networks for Pattern Recognition, Clarendon, UK (1995)Google Scholar
- 5.Principe, J.C., Euliano, N.R., Lefebvre, W.C.: Neural and Adaptive Systems: Fundamentals through Simulations. John Wiley and Sons, Inc., New York (2000)Google Scholar
- 6.Rumelhart, D.E., McClelland, J.L. and the PDP Research Group.: Parallel Distributed Processing - Explorations in the Microstructure of Cognition, Volume I: Foundations. The MIT Press, Boston (1986)Google Scholar
- 8.Smith, S.K., Mandell, M.: A Comparison of Local Population Estimates: the Housing Unit Method vs. Component II, Ratio Correlation and Administrative Records. Journal of the American Statistical Association 79(386), 282–289 (1984)Google Scholar
- 9.Smith, S.K., Tayman, J., Swanson, D.A.: State and Local Population Projections Methodology and Analysis. Kluwer/Plenum, New York (2001)Google Scholar
- 11.US Bureau of the Census: Methodology for Estimates of State and County Total Popula-tion Available at, http://www.census.gov/population/methods/stco99.txt. Revised (April 2000)