Comparative Analysis of Machine Learning Techniques for Effective Anomaly Detection in Humidity and Temperature Data in WSNs

Wireless Sensor Networks (WSNs) are vital for monitoring environmental parameters like temperature and humidity. Accurate anomaly detection in sensor data is crucial for maintaining data reliability and application efficiency. This research conducts a comparative analysis of five machine learning techniques—Support Vector Machines (SVM), k-Nearest Neighbors (k-NN), Random Forests (RF), Naive Bayes (NB), and Long Short-Term Memory (LSTM)—to evaluate their effectiveness in detecting anomalies in WSN sensor readings. We analyze various performance indicators, encompassing accuracy, precision, recall, F1-score, and processing time, utilizing a labeled dataset from existing literature. Results indicate that while LSTM excels in accuracy with larger training samples, RF and k-NN consistently demonstrate strong performance across metrics. This study highlights the strengths and limitations of each technique, providing insights for the selection of suitable methods in anomaly detection within WSN applications.