Anomaly Detection for a Temperature Device Using Machine Learning: A Comparative Study of Algorithms
Nuzhat Noor Islam Prova · Research Square · 2024
Abstract This research paper delves into the use of machine learning techniques for anomaly detection in temperature readings from a device. The study employs a rich dataset comprising timestamped temperature values. Key methodologies include extensive feature engineering to enhance data interpretability and the application of K-Means clustering, Principal Component Analysis (PCA), EllipticEnvelope, and Isolation Forest algorithms. The primary objective is to discern intricate patterns and anomalies that may signify potential device malfunctions. Special attention is given to identifying temporal patterns, with a focus on analyzing anomalies during different times of the day and week. This paper highlights the effectiveness of these algorithms in detecting anomalies and offers insights into their comparative performance. The findings have significant implications for predictive maintenance and operational efficiency in temperature-sensitive devices.