Predictive Data Analytics Using Machine Learning for IoT Sensor Network

• Published in: SN computer science Vol. 6; no. 7; p. 887
• Main Authors: Rajesh, Chawla, Mridul
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 11.10.2025; Springer Nature B.V
• Subjects: Accuracy; Artificial intelligence; Computer Imaging; Computer Science; Computer Systems Organization and Communication Networks; Data Structures and Information Theory; Decision making; Genetic algorithms; Information Systems and Communication Service; Internet of Things; Machine learning; Microelectromechanical systems; Original Research; Particle swarm optimization; Pattern Recognition and Graphics; Real time; Resource management; Sensor arrays; Sensors; Software Engineering/Programming and Operating Systems; Vision; Wireless sensor networks; Wireless sensor networks; Machine learning (ML); IoT sensor networks; Predictive data analytics (PDA); Naïve bayes classifier; Time-Series forecasting
• ISSN: 2661-8907; 2662-995X; 2661-8907
• DOI: 10.1007/s42979-025-04428-w

Predictive data analytics (PDA) and Machine learning (ML) into IoT sensor networks facilitates the generation of real-time insights and enables proactive decision-making. By analyzing vast arrays of sensor data such as temperature, humidity, and motion these technologies enhance operational efficiency, predict equipment failures, and identify anomalies. Advances in wireless sensor networks and micro-electromechanical systems have made these networks more practical and valuable, particularly in manufacturing, healthcare, and smart cities. Employing sophisticated algorithms like ARIMA for time-series forecasting and OMLEA for model optimization improves forecast accuracy and resource management. Effective data preparation using methods such as Genetic Algorithms and Particle Swarm Optimization (PSO) ensures the use of high-quality data for model training. Among various machine learning models, the Naive Bayes classifier excels in IoT sensor network applications, achieving an impressive 99.86% accuracy and a 99.91% F1-score, outperforming other models in recall and balance between real positives and false positives/negatives. The results demonstrated the potential of ML and PA to drive innovation and efficiency in IoT environments, despite challenges related to data quality, scalability, and privacy.