Optimized Decision-Making Framework for Detecting Important Factors Influencing Students’ Innovative Capabilities

Abstract

Developing innovative capabilities in university students is essential for individual career success and broader societal advancement. This study introduces a predictive Feature Selection (FS) model named bWRBA-SVM-FS, which combines an enhanced Bat Algorithm (BA) and Support Vector Machine (SVM). To enhance the optimization capability of BA, water follow search and random follow search are introduced to optimize the efficiency and accuracy of the feature subset search. Experimental validation conducted on the IEEE CEC 2017 benchmark functions and the talented innovative capacity dataset demonstrates the efficacy of the proposed method relative to peer and prominent machine learning models. The experimental results reveal that the predictive accuracy of the bWRBA-SVM-FS model is 97.503%, with a sensitivity of 98.391%. Our findings indicate significant predictors of innovation capacity, including project application goals, educational background, and interdisciplinary thinking abilities. The bWRBA-SVM-FS model offers effective strategies for talent selection in higher education, fostering the development of future research leaders.