On Machine Learning with Imbalanced Data and Research Quality Evaluation Methodologies

In this article a synoptic review of machine learning techniques with imbalanced data and a class of corresponding learning algorithms is presented. This class of algorithms includes the meta-algorithms: Cost sensitive, Metacost, Rotation forest-cost sensitive, rotation forest-smote. Four learning algorithms (with base classifiers J48 and part processing with F-measure and a predetermined imbalanced data set) are compared in the computational environment WEKA leading to comparative numerical results. The basic concepts of research quality evaluation methodologies are presented, an adaptive citation qualitative-quantitative approach and advanced bibliometric indicators are given. Basic components of research quality performance such as research journal cited publications, citing publications and research quality evaluations at various academic levels are considered and corresponding numerical results are given. An alternative approach using certain machine learning algorithms with imbalanced data in the case of research quality evaluation methodologies is proposed.