A Subspace-based Approach for Dimensionality Reduction and Important Variable Selection
This work addresses the problem of interpretable analysis in high-dimensional systems, such as failure prediction in materials, but it is incremental as it builds on existing dimensionality reduction and variable selection techniques.
The paper tackles the challenge of interpretable dimensionality reduction and variable selection in high-dimensional data by proposing a subspace-based method using randomized search and ensemble models. It demonstrates improved prediction and variable selection performance on failure prediction data for composite/metal hybrid structures, outperforming existing alternatives.
An analysis of high-dimensional data can offer a detailed description of a system but is often challenged by the curse of dimensionality. General dimensionality reduction techniques can alleviate such difficulty by extracting a few important features, but they are limited due to the lack of interpretability and connectivity to actual decision making associated with each physical variable. Variable selection techniques, as an alternative, can maintain the interpretability, but they often involve a greedy search that is susceptible to failure in capturing important interactions or a metaheuristic search that requires extensive computations. This research proposes a new method that produces subspaces, reduced-dimensional physical spaces, based on a randomized search and leverages an ensemble of critical subspace-based models, achieving dimensionality reduction and variable selection. When applied to high-dimensional data collected from the failure prediction of a composite/metal hybrid structure exhibiting complex progressive damage failure under loading, the proposed method outperforms the existing and potential alternatives in prediction and important variable selection.