An Empirical Study on the Impact of Dataset Characteristics and Hyperparameters on Machine Learning Model Performance

Sony Annem

Abstract: In this experiment we evaluated the performance of Naive Bayes, Support Vector Machine (classification and regression models), we tried to plot the decision boundary for linear kernel and rbf kernel and used Ensemble Learning methods such as Bagging and AdaBoost, worked on three types of datasets (.mat file format?s) such as Spam, Voting and Volcanoes.

Keywords: Python, Jupyter Notebook, Classification algorithms (Decision tree, Naive Bayes, Support Vector Machine), Support Vector Regression, Ensemble methods (bagging, boosting)