Internet-of-Things-Based Rain Detection Device Using Machine Learning Classification for Rain Prediction at Telkom University

Erratic weather changes over time have a significant impact. Unpredictable weather changes can bring up various problems, one of which is hampering the activities of the Telkom University civitas in outdoor activities. However, there is still a need for weather information. Due to the need for more information about weather predictions in the campus environment, Internet of Things (IoT) technology was proposed as a solution for providing weather detection information. This IoT tool has a variety of sensors that can be used for weather detection, such as rain sensors, temperature, humidity, light, and air pressure. The results from the sensors will then be classified using machine learning algorithms. Machine learning is used to detect weather changes and obtain information about them. The data obtained from the IoT utility will be labeled and categorized into three categories: sunny, drizzle, and rain, based on specific parameters. This study compared the prediction performance of four machine learning algorithms: Logistic Regression, Support Vector Machine (SVM), XGBoost, and Artificial Neural Network (ANN). The results demonstrate that XGBoost produces exceptional results with a 100% accuracy rate and 100% precision, while the others got around 99.20% to 99.31% and 92.74 to 97.62%. The results indicate that the IoT-based weather detection algorithm XGBoost can effectively predict the weather.