Probabilistic Forecast
Probabilistic forecasting aims to predict future events not as single points but as probability distributions, quantifying uncertainty inherent in predictions. Current research emphasizes improving the accuracy and calibration of these forecasts across diverse domains, focusing on advanced model architectures like neural networks (including variations such as LSTMs, VAEs, and diffusion models), Gaussian processes, and ensemble methods, often incorporating techniques like flow matching and conformal inference. These advancements have significant implications for various fields, enhancing decision-making in areas such as energy management, finance, and healthcare by providing more reliable and informative predictions under uncertainty.
Papers
When Rigidity Hurts: Soft Consistency Regularization for Probabilistic Hierarchical Time Series Forecasting
Harshavardhan Kamarthi, Lingkai Kong, Alexander Rodríguez, Chao Zhang, B. Aditya Prakash
Large Language Model Prediction Capabilities: Evidence from a Real-World Forecasting Tournament
Philipp Schoenegger, Peter S. Park