DoFlow: Causal Generative Flows for Interventional and Counterfactual Time-Series Prediction
This work addresses the need for causal reasoning in time-series forecasting for domains like hydropower and healthcare, representing a novel method for a known bottleneck rather than a foundational advance.
The authors tackled the problem of causal forecasting for multivariate time-series under interventional and counterfactual queries by introducing DoFlow, a flow-based generative model over a causal DAG, which achieved accurate observational predictions and enabled causal forecasting and anomaly detection in synthetic and real-world datasets.
Time-series forecasting increasingly demands not only accurate observational predictions but also causal forecasting under interventional and counterfactual queries in multivariate systems. We present DoFlow, a flow based generative model defined over a causal DAG that delivers coherent observational and interventional predictions, as well as counterfactuals through the natural encoding and decoding mechanism of continuous normalizing flows (CNFs). We also provide a supporting counterfactual recovery result under certain assumptions. Beyond forecasting, DoFlow provides explicit likelihoods of future trajectories, enabling principled anomaly detection. Experiments on synthetic datasets with various causal DAG and real world hydropower and cancer treatment time series show that DoFlow achieves accurate system-wide observational forecasting, enables causal forecasting over interventional and counterfactual queries, and effectively detects anomalies. This work contributes to the broader goal of unifying causal reasoning and generative modeling for complex dynamical systems.