Assessing a Mixture Model for Clustering with the Integrated Completed Likelihood
IEEE Transactions on Pattern Analysis and Machine Intelligence
Editorial: Special issue on variable selection and robust procedures
Computational Statistics & Data Analysis
Hi-index | 0.03 |
This paper formulates a mixture model for modeling unobserved heterogeneity of explanatory mechanism. Our model allows for different sets of regressors and/or different interactions among the same regressors in different regression regimes. The model is demonstrated with particular interest to the censored dependent variable. A two-step procedure is proposed for model identification. The first step is to identify the number of regression regimes with each regime, including all regressors. The second step is to select regressors in the regression regimes. The results of our simulation studies suggest that the procedure works well. Two microeconometric applications are provided.