Copula Conformal Prediction for Multi-step Time Series Forecasting
This work addresses uncertainty measurement for robust machine learning systems in time series forecasting, offering a novel method for multi-step predictions, though it is incremental in extending conformal prediction to handle temporal dependencies.
The paper tackles the problem of uncertainty quantification in multi-step time series forecasting by proposing CopulaCPTS, a copula-based conformal prediction algorithm that accounts for temporal dependencies, and shows it produces more calibrated and sharp confidence intervals than existing techniques on synthetic and real-world datasets.
Accurate uncertainty measurement is a key step to building robust and reliable machine learning systems. Conformal prediction is a distribution-free uncertainty quantification algorithm popular for its ease of implementation, statistical coverage guarantees, and versatility for underlying forecasters. However, existing conformal prediction algorithms for time series are limited to single-step prediction without considering the temporal dependency. In this paper, we propose a Copula Conformal Prediction algorithm for multivariate, multi-step Time Series forecasting, CopulaCPTS. We prove that CopulaCPTS has finite sample validity guarantee. On several synthetic and real-world multivariate time series datasets, we show that CopulaCPTS produces more calibrated and sharp confidence intervals for multi-step prediction tasks than existing techniques.