This tree-based method deals with high dimensional longitudinal data with correlated features through the use of a piecewise random effect model. FREE tree also exploits the network structure of the features, by first clustering them using Weighted Gene Co-expression Network Analysis ('WGCNA'). It then conducts a screening step within each cluster of features and a selecting step among the surviving features, which provides a relatively unbiased way to do feature selection. By using dominant principle components as regression variables at each leaf and the original features as splitting variables at splitting nodes, FREE tree delivers easily interpretable results while improving computational efficiency.
FREEtree, a tree-based method for high dimensional longitudinal data with correlated features. 'FREEtree' deals with longitudinal data by using a piecewise random effect model. It also exploits the network structure of the features, by first clustering them using Weighted Gene Co-expression Network Analysis ('WGCNA'). It then conducts a screening step within each cluster of features and a selecting step among the surviving features, which provides a relatively unbiased way to do feature selection. By using dominant principle components as regression variables at each leaf and the original features as splitting variables at splitting nodes, 'FREEtree' maintains 'interpretability' and improves computational efficiency.
This project is licensed under GPL-3