Objective Bayesian transformation and variable selection using default Bayes factors
- 112 Downloads
In this work, the problem of transformation and simultaneous variable selection is thoroughly treated via objective Bayesian approaches by the use of default Bayes factor variants. Four uniparametric families of transformations (Box–Cox, Modulus, Yeo-Johnson and Dual), denoted by T, are evaluated and compared. The subjective prior elicitation for the transformation parameter \(\lambda _T\), for each T, is not a straightforward task. Additionally, little prior information for \(\lambda _T\) is expected to be available, and therefore, an objective method is required. The intrinsic Bayes factors and the fractional Bayes factors allow us to incorporate default improper priors for \(\lambda _T\). We study the behaviour of each approach using a simulated reference example as well as two real-life examples.
KeywordsBayesian model selection Fractional Bayes factor Intrinsic Bayes factor Posterior model probabilities Transformation family selection Variable selection
- Berger, J.O., Pericchi, L.R.: The intrinsic Bayes factor for linear models. In: Bernardo, J., Berger, J., Dawid, A., Smith, A. (eds.) Bayesian Statistics, vol. 5, pp. 25–44. Oxford University Press, Oxford (1996a)Google Scholar
- Berger, J.O., Pericchi, L.R.: Objective Bayesian methods for model selection: Introduction and comparison. In: Lahiri, P. (ed.) ‘Model selection’, Lecture Notes-Monograph Series, vol. 38, Institute of Mathematical Statistics, pp. 135–207 (2001)Google Scholar
- Kim, S., Chen, M.H., Ibrahim, J.G., Shah, A.K., Lin, J.: Bayesian inference for multivariate meta-analysis Box–Cox transformation models for individual patient data with applications to evaluation of cholesterol-lowering drugs. Stat. Med. 32, 3972–3990 (2013)Google Scholar
- Thall, P.F., Russell, K.E., Simon, R.M.: Variable selection in regression via repeated data splitting. J. Am. Stat. Assoc. 6, 416–434 (1997)Google Scholar