This study deals with tree-based techniques and functional data analysis (FDA) [1] for supervised classification of curves representing high-dimensional biomedical data recorded over time. Recently [2] proposed Functional Classification Trees (FCTs) and Functional Random Forest (FRF) [3] using b-spline representation and the Functional Principal Components Decomposition (FPCD) as possible basis transformation to obtain features from curves for training the classifiers. In our proposal, an original contribution is also given by new interpretative tools of the functional classification rules in the functional framework . Applications on ECG data have shown the effectiveness of the proposed functional classifiers in terms of accuracy and their usefulness in terms of interpretability.
Functional Random Forest for Biomedical Signals Classification and Interpretative Tools
Fabrizio, Maturo;
2022-01-01
Abstract
This study deals with tree-based techniques and functional data analysis (FDA) [1] for supervised classification of curves representing high-dimensional biomedical data recorded over time. Recently [2] proposed Functional Classification Trees (FCTs) and Functional Random Forest (FRF) [3] using b-spline representation and the Functional Principal Components Decomposition (FPCD) as possible basis transformation to obtain features from curves for training the classifiers. In our proposal, an original contribution is also given by new interpretative tools of the functional classification rules in the functional framework . Applications on ECG data have shown the effectiveness of the proposed functional classifiers in terms of accuracy and their usefulness in terms of interpretability.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
