Events

Global AI Frontier Lab Seminar Series: Causal Inference with Time-Series Panel Data

Speaker: Saeyoung Rho (Columbia University)

Location: 1 MetroTech Center, Room Global AI Frontier Lab (Floor 22)

Date: Monday, October 26, 2026

We are pleased to announce the next session of the Global AI Frontier Lab: Seminar Series on October 26th, 2026. Saeyoung Rho will be presenting “Causal Inference with Time-Series Panel Data". Dinner & networking will begin at 6:00 PM and the seminar will start at 7:00 PM EST. The seminar will be held at the Global AI Frontier Lab at 1 Metrotech Center, Brooklyn, NY 11201. This event will be in-person & online. In-person attendance is strongly encouraged for Lab researchers in NYC. All attendees must RSVP to participate. For online attendees, a Zoom link will be sent out prior to the event. Please reach out to global-ai-frontier-lab@nyu.edu with any questions. We hope to see you there!

Title: Causal Inference with Time-Series Panel Data

Abstract: A central question in causal inference is, “What would have happened if we hadn’t intervened?” Answering it requires counterfactual reasoning about a world we never observe. Synthetic control is one of the most popular tools for doing this with observational time-series panel data. Suppose California adopts a new policy that other states do not. Synthetic control estimates California’s counterfactual trajectory as a weighted combination of the other states’ time series. Since economists Abadie and Gardeazabal introduced it in 2003, the method has spread across statistics, economics, operations, and computer science. 

The original method, however, rests on strong assumptions about how the data are generated, and it was designed for relatively small datasets. Modern panel data look very different: many more units, longer and noisier time series, and richer temporal dynamics. In this talk, I’ll show how machine learning can bring synthetic control into this regime. I’ll present two methods: Cluster Synthetic Control (ClusterSC), which adds a donor-selection step to handle a large pool of units, and Time-Aware Synthetic Control (TASC), which models the data with a state-space model to capture temporal dynamics. I’ll close with our recent work on generating realistic synthetic-control-style panel datasets, which opens the door to systematically benchmarking synthetic control methods.

Bio: Saeyoung Rho is a computer scientist developing scalable and privacy-preserving methods for personalized causal inference to support the design of policies, algorithms, and AI services. She is a final-year PhD candidate at Columbia University, co-advised by Professors Vishal Misra and Rachel Cummings. During her PhD, she gained industry experience at Amazon AWS and Google X. Prior to her PhD, she studied at Massachusetts Institute of Technology (MIT) and Korea Advanced Institute of Science and Technology (KAIST), earning a Master’s in Electrical Engineering and Computer Science and a Master’s in Technology and Policy from MIT, as well as both a Master’s and a Bachelor’s in Industrial Design from KAIST.

Webpage: saeyoungrho.com