Performance of Machine Learning Algorithms with Different K Values in K-fold CrossValidation

The numerical value of k in a k-fold cross-validation training technique machine learning predictive models is an essential element that impacts the model’s performance. A right choice results better accuracy, while poorly chosen for might affect In literature, most commonly used values are five (5) or ten (10), as these two believed to give test error rate estimates suffer neither from extremely high bias nor very variance. However, there no formal rule. To best our knowledge, few experimental studies attempted investigate effect diverse different models. This paper empirically analyses prevalence and distinct (3, 5, 7, 10, 15 20) on validation performance four well-known algorithms (Gradient Boosting Machine (GBM), Logistic Regression (LR), Decision Tree (DT) K-Nearest Neighbours (KNN)). It was observed model differ one machine-learning algorithm another same classification task. empirical suggest = 7 offers slight increase validations accuracy area under curve measure with lesser computational complexity than 10 across MLA. We discuss detail study outcomes outline some guidelines beginners field selecting given