為了解決此問題，\citet{Zou2006} 提出稀疏主成分分析（Sparse PCA）的概念，將原本的降維問題轉換為回歸形式，進而達到變數選擇與結果解釋的目的。在此基礎上，本文進一步探討 Wang et al. (2025) 所提出的貝氏稀疏化主座標分析（Bayesian Sparse PCoA， BSPCoA），並結合由 \citet{Xia2002MAVE} 提出的最小平均變異數估計（MAVE）之非參數方法，以提升模型在降維同時的解釋能力。
最後，透過模擬實驗與實際資料分析，將所提出方法與傳統 PCoA 以及 BSPCoA 進行比較，驗證方法在理論與實務上的表現與優勢。;Principal Coordinates Analysis (PCoA) is a widely used dimension reduction method for analyzing similarities and differences in microbial community compositions. However， the traditional PCoA approach primarily captures the distance structure among samples and lacks interpretability， as it cannot identify which microorganisms (e.g.， fungi) contribute to the observed clustering patterns.
To address this limitation， \citet{Zou2006} proposed Sparse Principal Component Analysis (SPCA)， which reformulates the dimension reduction problem as a regression framework， enabling variable selection and improving interpretability. Building upon this idea， this study investigates the Bayesian Sparse Principal Coordinates Analysis (BSPCoA) proposed by \citet{Wang2025}， and further integrates it with the Minimum Average Variance Estimation (MAVE)， a nonparametric method， to enhance interpretability in the reduced space.
Finally， through simulation studies and real data analysis， the proposed approach is compared with traditional PCoA and BSPCoA to evaluate its performance， thereby validating its theoretical and practical advantages.
