So sánh phương pháp

	Phân tích tác động nhân quả tăng cường học máy	Nghiên cứu sự kiện bảng
Lĩnh vực	Suy luận nhân quả	Suy luận nhân quả
Họ	Regression model	Regression model
Năm ra đời	2015-2018	1990s–2020s (modern panel formulation)
Người khởi xướng	Brodersen et al. (foundational BSTS framework, 2015); Chernozhukov et al. (double ML augmentation, 2018)	Formalized by Freyaldenhoven, Hansen, Perez-Orive & Shapiro (2021); widely applied in finance (Fama et al. 1969) and policy evaluation
Loại	Quasi-experimental causal inference with ML	Quasi-experimental / causal panel design
Công trình gốc	Brodersen, K. H., Gallusser, F., Koehler, J., Remy, N., & Scott, S. L. (2015). Inferring causal impact using Bayesian structural time-series models. Annals of Applied Statistics, 9(1), 247-274.	Freyaldenhoven, S., Hansen, C., Perez-Orive, J., & Shapiro, J. M. (2021). Visualization, Identification, and Estimation in the Linear Panel Event-Study Design. NBER Working Paper 29170. National Bureau of Economic Research.
Tên gọi khác	ML-augmented causal impact, ML-CausalImpact, machine learning causal impact, ML-augmented BSTS	event-study regression, dynamic DiD, relative-time regression, distributed-lag panel model
Liên quan	6	4
Tóm tắt	Machine learning-augmented causal impact analysis combines quasi-experimental counterfactual reasoning with flexible ML prediction models to estimate the causal effect of an intervention on a time series outcome. Building on Brodersen et al.'s Bayesian structural time series (BSTS) framework and extended by double/debiased ML methods, it constructs a synthetic counterfactual from donor covariates and infers the treatment effect as the gap between observed and predicted post-intervention outcomes.	A panel event study estimates the dynamic causal effect of a treatment or policy by regressing an outcome on a full set of relative-time indicators — one for each period before and after the event — while controlling for unit and time fixed effects. The resulting coefficient plot shows how the treated units diverged from untreated units at each point in calendar time relative to their treatment date, making both pre-treatment trend violations and post-treatment effect trajectories immediately visible.