An Intelligent Anomaly Detection Approach for Accurate and Reliable Weather Forecasting at IoT Edges: A Case Study.

Area of Science:

• Environmental Science
• Computer Science
• Data Science

Background:

• Industrialization and urbanization negatively impact ecosystems, climate, and biodiversity.
• Rapid digitalization and insufficient data infrastructure exacerbate environmental challenges.
• Inaccurate Internet of Things (IoT) data compromises weather forecast reliability, hindering public safety and planning.

Purpose of the Study:

• To present an intelligent anomaly detection approach to address weather forecasting inaccuracies.
• To improve the accuracy and reliability of weather predictions in the face of urbanization and digitalization.
• To filter erroneous data at the edge of IoT for enhanced forecasting.

Main Methods:

• Implemented an intelligent anomaly detection approach for IoT data processing.
• Utilized five machine learning algorithms: Support Vector Classifier (SVC), Adaboost, Logistic Regression (LR), Naive Bayes (NB), and Random Forest (RF).
• Compared the performance of these algorithms in detecting anomalies in sensor data (time, temperature, pressure, humidity).

Main Results:

• The study compared anomaly detection metrics across five distinct machine learning algorithms.
• The proposed approach aims to filter missing, irrelevant, or anomalous data from sensor inputs.
• This filtering is crucial for enhancing the accuracy and reliability of weather predictions.

Conclusions:

• An intelligent anomaly detection method can mitigate weather forecasting issues caused by urbanization and digitalization.
• Edge data processing and anomaly filtering are vital for reliable weather predictions.
• Machine learning algorithms offer a robust framework for improving weather data quality and forecasting outcomes.