Spatio-Temporal Graphical Counterfactuals: An Overview

TL;DR

This paper surveys various counterfactual models and introduces a unified graphical causal framework for inferring spatio-temporal counterfactuals, addressing a gap in modeling spatial and temporal interactions.

Contribution

It provides a comprehensive comparison of existing counterfactual theories and proposes a novel unified graphical framework for spatio-temporal counterfactual inference.

Findings

Comparison of different counterfactual models and theories

Introduction of a unified graphical causal framework

Addresses the lack of graphical approaches for spatio-temporal data

Abstract

Counterfactual thinking is a crucial yet challenging topic for artificial intelligence to learn knowledge from data and ultimately improve performance for new scenarios. Many research works, including the Potential Outcome Model (POM) and the Structural Causal Model (SCM), have been proposed to address this. However, their modeling, theoretical foundations, and application approaches often differ. Moreover, there is a lack of graphical approaches for inferring spatio-temporal counterfactuals, that account for spatial and temporal interactions among multiple units. Thus, in this work, we aim to present a survey that compares and discusses different counterfactual models, theories and approaches. Additionally, we propose a unified graphical causal framework to infer spatio-temporal counterfactuals.