This research evaluates the impact of different choices of parameters in Singular Spectrum Analysis (SSA) for time series forecasting. The study compares methods for selecting SSA parameters and their effects on forecasting accuracy. Key highlights include:
The study uses real-world data sets related to atmospheric phenomena, analyzing the performance of SSA for short-term horizons. Results suggest that the method proposed by [15] for window length selection outperforms other methods evaluated. Automated grouping methods provided suboptimal results compared to manually selected prefix groupings. The research highlights the need for further development in automated grouping algorithms to enhance decision support tools' effectiveness.
In un'altra lingua
dal contenuto originale
arxiv.org
Approfondimenti chiave tratti da
by Teodor Knapi... alle arxiv.org 03-26-2024
https://arxiv.org/pdf/2403.16507.pdfDomande più approfondite