Hybrid KAN-XGBoost model improves electricity price forecasting

Researchers have developed a new hybrid framework combining Kolmogorov-Arnold Networks (KAN) with XGBoost for more accurate week-ahead electricity price forecasting in Australia's National Electricity Market. This approach leverages KAN's ability to model complex nonlinear relationships and XGBoost's robustness for capturing short-term price fluctuations. Experiments on real-world data showed the hybrid model significantly outperformed existing methods, reducing Mean Absolute Error by approximately 12% compared to standalone XGBoost and over 50% compared to a naive baseline. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a novel hybrid model that significantly improves electricity price forecasting accuracy, potentially aiding market participants in operational planning and risk management.

RANK_REASON Academic paper detailing a novel hybrid machine learning framework for a specific forecasting task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Kolmogorov-Arnold Network
• Long Short-Term Memory
• SARIMAX
• XGBoost

• paper
• other