Global Forecasting

Global forecasting aims to predict various phenomena across the globe, from weather patterns and air quality to famine risk and forest biomass. Current research heavily utilizes machine learning, employing diverse architectures like deep learning (including Transformers, CNNs, and RNNs), Random Forests, and Gaussian Processes, often integrated with data assimilation techniques to improve accuracy and incorporate real-time observations. These advancements enhance prediction capabilities for diverse applications, improving decision-making in areas such as disaster preparedness, resource management, and public health, while also driving methodological innovation in explainability and interpretability of complex models.

Papers

February 15, 2022

LIMREF: Local Interpretable Model Agnostic Rule-based Explanations for Forecasting, with an Application to Electricity Smart Meter Data
Dilini Rajapaksha, Christoph Bergmeir
Application Proficiency State of the Art Forecasting Smart Meter Local Interpretable Model Agnostic Explanation Consumption Forecasting Global Forecasting Interpretable Forecast

February 7, 2022

Gaussian Graphical Models as an Ensemble Method for Distributed Gaussian Processes
Hamed Jalali, Gjergji Kasneci
Gaussian Process Ensemble Method Conditional Distribution Global Forecasting Gaussian Graphical Model

November 13, 2021

LoMEF: A Framework to Produce Local Explanations for Global Model Time Series Forecasts
Dilini Rajapaksha, Christoph Bergmeir, Rob J Hyndman
New Framework Black Box Global Model Local Explanation Univariate Time Series Forecasting Global Forecasting

Global Forecasting

Papers

LIMREF: Local Interpretable Model Agnostic Rule-based Explanations for Forecasting, with an Application to Electricity Smart Meter Data

Gaussian Graphical Models as an Ensemble Method for Distributed Gaussian Processes

LoMEF: A Framework to Produce Local Explanations for Global Model Time Series Forecasts