Classification in MerQur: From ROC Curves to Gradient Boosting

Abstract

Classification is one of the most common tasks in academic research and applied data science: assigning an observation to one of predefined categories. On the bridge between classical statistics (logistic regression) and modern machine learning (random forest, gradient boosting, SVM), the correct measurement and reporting of classification performance requires special methodological care. This study introduces in detail the 6 analyses offered in MerQur’s Classification category: ROC Curve, TSS (True Skill Statistic), Confusion Matrix Metrics, Random Forest Classification, Support Vector Machine (SVM), and Gradient Boosting Classification. For each: (i) the basis of the method and its place in the classification task, (ii) hyperparameters and selection strategies, (iii) form fields and options in MerQur, (iv) reported performance metrics (accuracy, precision, recall, F1, AUC, TSS, kappa, Matthews correlation), and (v) interpretation guidance for a typical research question. The role of ROC and AUC in threshold-independent evaluation, the selection of correct metrics in imbalanced classes (F1 / MCC / TSS), the interpretation of variable importance in Random Forest, kernel selection in SVM, and overfitting control in Gradient Boosting are discussed. MerQur’s Classification category covers the spectrum from classical diagnostic threshold analysis to modern ensemble methods within a single graphical interface and includes k-fold cross-validation as standard.