Research map / Learning theory and algorithms / Combinatorial and search optimization
Synthetic Design: An Optimization Approach to Experimental Design with Synthetic Controls
Nick Doudchenko, Khashayar Khosravi, Jean Pouget-Abadie, Sebastien Lahaie, Miles Lubin, Vahab Mirrokni, Jann Spiess, Guido Imbens
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| optimize treated units with per-unit design instead of random synthetic control | synthetic control · BLS simulation, heterogeneous, K=3 | unit-level RMSE (x10^3): 15.9 -> 13.9 | Improved |
| optimize treated units with per-unit design instead of random synthetic control | synthetic control · BLS simulation, heterogeneous, K=7 | unit-level RMSE (x10^3): 19.0 -> 16.0 | Improved |
| per-unit design instead of random synthetic control | synthetic control · BLS simulation, homogeneous, K=3 | ATET RMSE (x10^3): 9.7 -> 8.5 | Improved |
| per-unit design instead of random synthetic control | synthetic control · BLS simulation, homogeneous, K=7 | ATET RMSE (x10^3): 10.3 -> 8.3 | Improved |
| synthetic control with random treatment instead of random diff-in-means | diff-in-means · BLS simulation, homogeneous, K=3 | ATET RMSE (x10^3): 12.1 -> 9.7 | Improved |
| two-way global design instead of diff-in-means | diff-in-means · BLS simulation, homogeneous, K=3 | ATET RMSE (x10^3): 12.1 -> 8.4 | Improved |
| two-way global design instead of per-unit design | per-unit · BLS simulation, homogeneous, K=3 | unit-level RMSE (x10^3): 13.9 -> 8.4 | Improved |
| two-way global design instead of per-unit design | per-unit · BLS simulation, heterogeneous, K=3 | unit-level RMSE (x10^3): 13.9 -> 27.6 | Hurt |
| two-way global design instead of per-unit design | per-unit · BLS simulation, heterogeneous, K=7 | unit-level RMSE (x10^3): 16.0 -> 32.5 | Hurt |
| one-way instead of two-way global design | two-way global · BLS simulation, homogeneous, K=3 | ATET RMSE (x10^3): 8.4 -> 8.5 | No clear effect |
| two-way global design in heterogeneous setting vs per-unit | per-unit · BLS simulation, heterogeneous, K=7 | ATET RMSE (x10^3): 8.3 -> 8.9 | Hurt |
| increase treated units from K=3 to K=7 for synthetic control | synthetic control · BLS simulation, homogeneous | unit-level RMSE (x10^3): 15.9 -> 19.0 | Hurt |
Closest papers by meaning
- Data-driven Design of Randomized Control Trials with Guaranteed Treatment Effects (ICML 2025 · similarity 0.83)
- Active Adaptive Experimental Design for Treatment Effect Estimation with Covariate Choice (ICML 2024 · similarity 0.83)
- Efficient Randomized Experiments Using Foundation Models (NeurIPS 2025 · similarity 0.83)
- Integer Programming for Generalized Causal Bootstrap Designs (ICML 2025 · similarity 0.83)
- Stronger Neyman Regret Guarantees for Adaptive Experimental Design (ICML 2025 · similarity 0.83)
- Targeted Synthetic Control Method (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
