A reproducible world-trade model connecting network structure, strategic sectors, and explicit probabilistic forecasts.
The operating problem
A trade network can look resilient in aggregate while depending heavily on a few countries and strategic sectors. Leaders need to see the assumptions behind a forecast.
What I built
Built a Python replication package, network analysis pipeline, validation checks, and ten probabilistic forecasts. Published the method, findings, and a correction to the model’s original confidence estimates.
The leadership connection
Decision science, model validation, uncertainty communication, and the willingness to correct an attractive result when the evidence changes.
Evidence & boundaries
Original public research; MIT-licensed replication package. Forecast outcomes remain unresolved. The inline scenario uses synthetic data; it is not a result from the research model.
- Python
- NetworkX
- Monte Carlo
- Bayesian inference
- Probabilistic forecasting