Research
Research interests: time series, high-dimensional regression, and post-selection inference — especially when dependence and selection make classical statements fragile.
In practice we rarely fix a model before looking at the data. Once selection has happened, intervals and p-values computed as if the model were given can be badly wrong. My work asks how to keep inferential claims honest under selection and temporal dependence.
Post-selection estimators and valid inference for time-series models · 2021–2023 · NSFC 12001461 · RMB 240K · completed.
The project focused on estimators and inferential procedures that remain meaningful after model selection in time-series settings — a bridge between methodological research and the claims we ask students to defend.
Co-Principal Investigator · 2025–2027 · 1+1+1 Joint Collaboration Fund · RMB 255K · ongoing (2025A0505000070).
This collaboration sits next to the teaching appointment: can adaptive systems help students practice judgment — choosing models, reading uncertainty, noticing when a claim is too strong — rather than only accelerating content delivery?
Co-PI on a completed industry collaboration with Guotai Junan Securities on machine-learning based intelligent routing research (2025).