An Integrated Approach for Screening Multidimensional Indicators in Combat Capability Assessment

Addressing the unique challenges of multidimensional correlations, multicollinearity, data scarcity, and irregularity in combat capability assessment, this study proposes a tailored integrated qualitative–quantitative approach for efficient indicator screening. Unlike standalone or generic applications of dimensionality reduction techniques, the approach intentionally combines principal component analysis (PCA), factor analysis (FA), and gray relational analysis (GRA) to align with the specific demands of combat contexts. PCA first reduces dimensionality by transforming correlated variables into orthogonal principal components, mitigating multicollinearity while retaining key variance; FA then refines this reduced set by extracting interpretable latent factors and clarifying their practical meaning via rotation, ensuring alignment with real-world operational scenarios; finally, GRA—selected for its robustness to small/irregular datasets—is introduced as an auxiliary tool to validate and optimize the indicator set. Compared to traditional or generic statistical methods, this integration achieves three key improvements: (1) it simultaneously resolves multicollinearity and data scarcity, two long-standing dilemmas unique to combat capability assessments; (2) it balances statistical rigor with operational interpretability; and (3) it reduces indicator sets by 54% while retaining over 80% of critical combat-related information, significantly enhancing assessment efficiency.