A Bayesian Nonparametric Learning Approach to Ensemble Models Using the Proper Bayesian Bootstrap

Bootstrap resampling techniques, introduced by Efron and Rubin, can be presented in a general Bayesian framework, approximating the statistical distribution of functional ?(F), where F is random function. Efron’s Rubin’s bootstrap procedures extended, introducing an informative prior through Proper bootstrap. In this paper different techniques are used compared predictive classification regression models based on ensemble approaches, i.e., bagging involving decision trees. bootstrap, proposed Muliere Secchi, to sample posterior over trees, distributions covariates target variable. The results obtained with respect other competitive employing techniques. empirical analysis reports simulated real data.