DCMD: Distance-based classification using mixture distributions on microbiome data

Current advances in next-generation sequencing techniques have allowed researchers to conduct comprehensive research on the microbiome and human diseases, with recent studies identifying associations between health outcomes for a number of chronic conditions. However, data structure, characterized by sparsity skewness, presents challenges building effective classifiers. To address this, we present an innovative approach distance-based classification using mixture distributions (DCMD). The method aims improve performance community data, where predictors are composed sparse heterogeneous count data. This models inherent uncertainty counts estimating distribution sample representing each observation as distribution, conditional observed estimated mixture, which then used inputs classification. is implemented into k -means -nearest neighbours framework. We develop two distance metrics that produce optimal results. model assessed simulated study results compared against existing machine learning approaches. proposed competitive when other approaches, shows clear improvement over commonly classifiers, underscoring importance modelling achieving range applicability robustness make viable alternative source code available at https : //github . com/kshestop/DCMD academic use.