Industrial Equipment Monitoring Dataset for Predictive Maintenance Analysis

Abstract

This study develops and evaluates a Support Vector Machine (SVM) model using a Radial Basis Function (RBF) kernel to detect faulty conditions in systems based on sensor data (temperature, pressure, vibration, humidity). The data is processed through normalization and split into training and testing sets. The evaluation results show an overall model accuracy of 0.93. The model is highly effective in identifying normal conditions (precision 0.93, recall 1.00), but less optimal in detecting faulty conditions (precision 0.96, recall 0.30), indicating a high number of false negatives and a low F1-score (0.45) for this class. The ROC AUC score of 0.892 indicates good overall discriminative ability. This performance gap is likely due to class imbalance. Enhancing faulty detection through class imbalance handling or further model optimization is recommended for critical applications.