Model population analysis for variable selection

Abstract

aTo build a credible model for given chemical or biological or clinical data, it may be helpful to first get somewhat better insight into the data itself before modeling and then to present the statistically stable results derived from a large number of sub-models established only on one dataset with the aid of Monte Carlo Sampling (MCS). In the present work, a concept model population analysis (MPA) is developed. Briefly, MPA could be considered as a general framework for developing new methods by statistically analyzing some interesting parameters (regression coefficients, prediction errors, etc.) of a number of sub-models. New methods are expected to be developed by making full use of the interesting parameter in a novel manner. In this work, the elements of MPA are first considered and described. Then, the applications for variable selection and model assessment are emphasized with the help of MPA.

Model Population Analysis for Statistical Model Comparison

“…Its extended version, Monte Carlo UVE (MCUVE), was recently proposed [27,31]. Mimicking the principle of "survival of the fittest" in Darwin's evolution theory, we developed a variable selection method in our previous work, called competitive adaptive reweighted sampling (CARS) [8,28,32,33], which was shown to have the potential to identify an optimal subset of variables that show high predictive performances. The source codes of CARS are freely available at [34,35].…”

“…Therefore, we have reasons to say that this kind of comparison is lack of statistical assessment and also at the risk of drawing wrong conclusions. We recently proposed model population analysis (MPA) as a general framework for designing chemometrics/bioinformatics methods [8]. MPA has been shown to be promising in outlier detection and variable selection.…”

GC–MS Based Serum Metabolomic Analysis of Isoflurane-Induced Postoperative Cognitive Dysfunctional Rats: Biomarker Screening and Insight into Possible Pathogenesis

“…In the analysis of metabolites, the existence of some uninformative and/or interfering variables is unavoidable because of the complexity of data system. Furthermore, these uninformative and/or interfering perturbances are very liable to cover up the important metabolic perturbance associated with pathological process, and subsequently make it difficult to discriminate abnormal from normal systems and to identify the informative biomarkers [10]. Hence, it seems to be necessary to eliminate the uninformative and interfering metabolites and to pick out the key ones by variable selection process.…”

Recipe for revealing informative metabolites based on model population analysis

“…Aimed at taking into account the synergetic effect among multiple variables and Motivated by Breiman's work, we in the present study introduce a novel method, named subwindow permutation analysis (SPA), by incorporating the Monte Carlo technique and strictly implementing the idea of model population analysis (MPA) (Li et al 2009a). Briefly, MPA is a newly proposed strategy for developing data analysis methods based on Monte Carlo sampling (MCS).…”