DeepTSF: Codeless machine learning operations for time series forecasting

TL;DR

DeepTSF addresses the need for accessible, reproducible MLOps in time-series forecasting by delivering a codeless, end-to-end framework that automates data loading, preprocessing, model training, evaluation, and deployment. It combines a React-based front-end, a Python/Javascript back-end, Docker-based deployment, and Dagster orchestration with MLflow tooling to enable multi-horizon forecasting while enforcing secure access via OpenID Connect. Key contributions include a modular architecture with MLflow-based tracking/registry, SHAP explanations, Optuna-driven hyperparameter tuning, and a dual UI/CLI interface that supports both non-programmatic use and expert customization. The approach is validated in the I-NERGY project for DL-based electricity load forecasting, demonstrating practical impact in energy systems and offering a path to broader applicability across time-series domains.

Abstract

This paper presents DeepTSF, a comprehensive machine learning operations (MLOps) framework aiming to innovate time series forecasting through workflow automation and codeless modeling. DeepTSF automates key aspects of the ML lifecycle, making it an ideal tool for data scientists and MLops engineers engaged in machine learning (ML) and deep learning (DL)-based forecasting. DeepTSF empowers users with a robust and user-friendly solution, while it is designed to seamlessly integrate with existing data analysis workflows, providing enhanced productivity and compatibility. The framework offers a front-end user interface (UI) suitable for data scientists, as well as other higher-level stakeholders, enabling comprehensive understanding through insightful visualizations and evaluation metrics. DeepTSF also prioritizes security through identity management and access authorization mechanisms. The application of DeepTSF in real-life use cases of the I-NERGY project has already proven DeepTSF's efficacy in DL-based load forecasting, showcasing its significant added value in the electrical power and energy systems domain.

Figures (7)

• Figure 1: The DeepTSF architecture
• Figure 2: Introductory web pages of DeepTSF
• Figure 3: DeepTSF codeless forecast page.
• Figure 4: DeepTSF experiment tracking page.
• Figure 5: DeepTSF experiment tracking and evaluation page results.