Peterson, L.E., Coleman, M.A. Logistic Ensembles for Principal Direction and Random Spherical Linear Oracles (343K, pdf) ABSTRACT. Principal direction linear oracle (PDLO) and random spherical linear oracle (RSLO) ensemble classifiers for DNA microarray gene expression data are proposed. The oracle assigns different training(testing) samples to 2 sub-classifiers of the same type using hyperplane splits in order to increase the diversity of voting results since errors are not shared across sub-classifiers. Eleven classifiers were evaluated for performance as the base classifier including k nearest neighbor (kNN), naive Bayes classifier (NBC), linear discriminant analysis (LDA), learning vector quantization (LVQ1), polytomous logistic regression (PLOG), artificial neural networks (ANN), constricted particle swarm optimization (CPSO), kernel regression (KREG), radial basis function networks (RBFN), gradient descent support vector machines (SVMGD), and least squares support vector machines (SVMLS). Logistic ensembles (PLOG) resulted in the best performance when used as a base classifier for PDLO and RSLO. The performance of logistic ensembles based on random axis selection and rotation followed by hyperplane splits in PDLO increased with increasing CV-fold and iteration number. However, logistic ensembles employed with random hyperplane splits used in RSLO resulted in degeneration of performance at the greatest levels of CV-fold and iteration number.