Telescope: An Automated Hybrid Forecasting Approach on a Level-Playing Field

• DOI: 10.48550/arxiv.2309.15871
• Publication Date: 2023-01-01
• Authors: André Bauer; Mark Leznik; Michael Stenger; Robert Leppich; Nikolas Herbst; Samuel Kounev; Ian Foster
• Keywords: Pillar

In many areas of decision-making, forecasting is an essential pillar. Consequently, different methods have been proposed. From our experience, recently presented are computationally intensive, poorly automated, tailored to a particular data set, or they lack predictable time-to-result. To this end, we introduce Telescope, novel machine learning-based approach that automatically retrieves relevant information from given time series and splits it into parts, handling each them separately. contrast deep learning methods, doesn't require parameterization the need train fit multitude parameters. It operates with just one provides forecasts within seconds without any additional setup. Our experiments show Telescope outperforms recent by providing accurate reliable while making no assumptions about analyzed series.