Application of Kubernetes and Machine Learning in the Development of a Weather Forecasting System

Abstract

This paper presents the development of a scalable and automated system for real-time temperature prediction that integrates Kubernetes, Docker, and LSTM neural networks. The system architecture is based on a microservices model, including services for prediction, data management, visualization, and task scheduling. The implementation was carried out using Minikube Kubernetes clusters, with a PostgreSQL database for data storage and the Streamlit library for results visualitzation. The LSTM model was trained on meteorological data, enabling temperature forecasts for the next 24 hours and 7 days. The results demonstrate that the combination of Kubernetes and machine learning provides efficient orchestration and scalability in system design. This work establishes a foundation for further development toward cloud-based deployment and integration with automated data collection and processing systems.