Chengpiao Huang

By producing a full predictive demand distribution instead of just point forecasts, this model empowers managers to make informed pricing decisions that balance revenue and risk.

LLM-based digital twins can be made 50-90% more accurate with a lightweight calibration framework inspired by causal inference.