A comprehensive analysis of multi-view feature-set partitioning methods with machine learning algorithms
Ritika Singh, Vipin Kumar · 2025
Multi-view feature-set partitioning (MvFSP) is crucial in machine learning for enhancing classification accuracy by dividing datasets into multiple views. This study explores the effectiveness of different FSP methods with various machine learning algorithms to determine optimal combinations for diverse data types. Utilizing 11 machine learning algorithms, including Logistic Regression (LR), Decision Trees, Random Forest, Support Vector Machines (SVM), k-nearest Neighbors (KNN), Naive Bayes (NB), Gradient Boosting Machines (GBM), Adaboost, XGBoost, LightGBM, and Linear Discriminant Analysis (LDA) across six partitioning methods, namely Random Feature-set Partitioning (RFSP), Attribute Bagging (AB), Attribute Clustering (AC), Ferrer Diagram (FD), Bell Triangle (BT), and Music Rhythm Tree (MRT), the research evaluates their efficacy based on accuracy metrics. The results demonstrate that Ferrer Diagram and Bell Triangle methods consistently yield the highest accuracy gains. Moreover, boosting techniques and Logistic Regression benefit the most from FSP combinations. Statistical analyses, including Friedman&s;s ranking, highlight the significant impact of FSP methods on classification performance, emphasizing the need for personalized approaches based on dataset characteristics.