Adaptive Sampling for Probabilistic Forecasting under Distribution Shift
This addresses the challenge of making accurate probabilistic forecasts when real-world time series change due to external events like economic cycles or pandemics, though it appears incremental in approach.
The paper tackles the problem of time series forecasting under distribution shift by developing an adaptive sampling strategy that selects relevant historical time steps using Bayesian optimization. The method reduces forecasting error significantly for three out of five datasets compared to a base model.
The world is not static: This causes real-world time series to change over time through external, and potentially disruptive, events such as macroeconomic cycles or the COVID-19 pandemic. We present an adaptive sampling strategy that selects the part of the time series history that is relevant for forecasting. We achieve this by learning a discrete distribution over relevant time steps by Bayesian optimization. We instantiate this idea with a two-step method that is pre-trained with uniform sampling and then training a lightweight adaptive architecture with adaptive sampling. We show with synthetic and real-world experiments that this method adapts to distribution shift and significantly reduces the forecasting error of the base model for three out of five datasets.