Comparative Analysis of Feature Importance in Machine Learning Models for Predictive Maintenance

Grigorios Tzionis, Myrsini Ntemi, Ilias Gialampoukidis, Stefanos Vrochidis, Ioannis Kompatsiaris, Maro Vlachopoulou

Abstract: This paper explores the application of multiple machine learning models, including Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), Boosted Gradient (BG), and Classification and Regression Trees (CART), in the context of predictive maintenance for industrial machinery. Utilizing a comprehensive dataset specific to machine operations and maintenance requirements, we aim to identify the most effective model for predicting maintenance needs. Post-modeling, we employ SHapley Additive exPlanations (SHAP) and Permutation Feature Importance (PFI) to evaluate and compare the significance of various features in each model. This comparative analysis seeks to provide insights into the effectiveness of these feature importance techniques in the realm of predictive maintenance, thus contributing valuable knowledge to the field of industrial machine learning applications.

External IDs:doi:10.1007/978-3-031-93995-2_3