Comparison of Ensemble Approaches: Mixture of Experts and AdaBoost for a Regression Problem

Two machine learning approaches: mixture of experts and AdaBoost.R2 were adjusted to the real-world regression problem of predicting the prices of residential premises based on historical data of sales/purchase transactions. The computationally intensive experiments were conducted aimed to compare empirically the prediction accuracy of ensemble models generated by the methods. The analysis of the results was performed using statistical methodology including nonparametric tests followed by post-hoc procedures designed especially for multiple n×n comparisons. No statistically significant differences were observed among the best ensembles: two generated by mixture of experts and two by AdaBoost.R2 employing multilayer perceptrons and general linear models as base learning algorithms.