Temporal Predictive Multiplicity: Equally Accurate Time Series Models Yield Different Forecast Trajectories
To address this problem, we introduce temporal predictive multiplicity, a framework that characterizes disagreement over complete forecast trajectories among models with near-identical predictive performance.
ProofPaper ↗
Key points
- Models with near-identical predictive performance can yield substantially different predictions, a phenomenon known as predictive multiplicity.
- In time-series forecasting, however, predictions across horizons jointly define a trajectory, and horizon-wise comparisons can hide important differences in predictive behavior.
- We show that constraining predictive performance alone can still admit a broad range of different trajectories.
- Our framework, therefore, exposes a gap in existing multiplicity studies: models with indistinguishable predictive performance imply fundamentally different temporal trajectories, with consequential downstream effects.
Sources (1)
- [1]Temporal Predictive Multiplicity: Equally Accurate Time Series Models Yield Different Forecast TrajectoriesarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 12:53 PM
To address this problem, we introduce temporal predictive multiplicity, a framework that characterizes disagreement over complete forecast trajectories among models with near-identical predictive performance.
Models with near-identical predictive performance can yield substantially different predictions, a phenomenon known as predictive multiplicity.
Extractive summary: sentences quoted from the sources.