Interpretability and Causal Inference for Global Time Series Forecasting Methods

I gave a talk at the 2021 Facebook Forecasting Summit, covering the work of my PhD students Dilini Rajapaksha on Interpretability and Priscila Grecov on Causal Inference. See the following publications for details:

• Dilini Rajapaksha, Christoph Bergmeir, Rob J Hyndman (2023) LoMEF: A framework to produce local explanations for global model time series forecasts. In: International Journal of Forecasting, 39, (3), pp. 1424-1447.

• Dilini Rajapaksha, Christoph Bergmeir (2022) LIMREF: Local Interpretable Model Agnostic Rule-based Explanations for Forecasting, with an Application to Eletricity Smart Meter Data. In: Thirty-Sixth AAAI Conference on Artificial Intelligence (AAAI-22).

• Priscila Grecov, Kasun Bandara, Christoph Bergmeir, Klaus Ackermann, Sam Campbell, Deborah Scott, Dan Lubman (2021) Causal Inference Using Global Forecasting Models for Counterfactual Prediction. In: PAKDD 2021: Advances in Knowledge Discovery and Data Mining, Springer International Publishing, pp. 282-294.

• Priscila Grecov, Ankitha Nandipura Prasanna, Klaus Ackermann, Sam Campbell, Debbie Scott, Dan I Lubman, Christoph Bergmeir (2022) Probabilistic causal effect estimation with global neural network forecasting models. In: IEEE Transactions on Neural Networks and Learning Systems.