6. Modelling Conditional Densities Using Finite Smooth Mixtures

  1. Kerrie L. Mengersen4
  2. Christian P. Robert5
  3. D. Michael Titterington6
  1. Feng Li1,
  2. Mattias Villani2,
  3. Robert Kohn3

Published Online: 24 APR 2011

DOI: 10.1002/9781119995678.ch6

Book Title

Mixtures: Estimation and Applications

Mixtures: Estimation and Applications

Additional Information

How to Cite

Li, F., Villani, M. and Kohn, R. (2011) Modelling Conditional Densities Using Finite Smooth Mixtures, in Mixtures: Estimation and Applications (eds K. L. Mengersen, C. P. Robert and D. M. Titterington), John Wiley & Sons, Ltd, Chichester, UK. doi: 10.1002/9781119995678.ch6

Editor Information

  1. 4

    School of Mathematical Sciences, Queensland University of Technology, Australia

  2. 5

    Université Paris-Dauphine, CEREMADE, Paris, France

  3. 6

    University of Glasgow, Glasgow, UK

Author Information

  1. 1

    Department of Statistics, Stockholm University, Sweden

  2. 2

    Sveriges Riksbank, Stockholm, Sweden

  3. 3

    Australian School of Business, University of New South Wales, Sydney, Australia

Publication History

  1. Published Online: 24 APR 2011
  2. Published Print: 15 APR 2011

ISBN Information

Print ISBN: 9781119993896

Online ISBN: 9781119995678

SEARCH

CHAPTER TOOLS

Get PDF (489K)

Keywords:

  • modelling conditional densities - using finite smooth mixtures;
  • finite smooth mixtures, or mixtures of experts (ME) - knowing machine learning literature;
  • smooth mixtures, capable of approximating - large class of conditional distributions;
  • simple-and-many versus complex-and-few - modelling regression data-skewed response variable;
  • inference methodology - general MCMC scheme;
  • generalised linear model (GLM) - to variable selection case;
  • model comparison - components assumed known in MCMC scheme;
  • simulation study in Villani et al. (2009) - smooth mixture of homoscedastic Gaussian components for heteroscedastic data;
  • LIDAR data, first real dataset - using laser-emitted light to detect chemical compounds in atmosphere;
  • log predictive density score (LPDS) - fivefold cross-validation of electricity expenditure data

Summary

This chapter contains sections titled:

  • Introduction

  • The model and prior

  • Inference methodology

  • Applications

  • Conclusions

  • Acknowledgements

  • Appendix: Implementation details for the gamma and log-normal models

  • References

Get PDF (489K)