Stefano Blando.
Italiano

05 — Research

Trust

How we know: data, statistics and verification to establish which conclusions a model can support, comparing rival models and testing results under a change of method, data or policy.

A handful of simulation runs, or a single fitted model, is not yet evidence. This pillar combines data, robust and high-dimensional statistics, causal inference, statistical model checking and formal verification to attach quantified confidence to what models imply. It compares rival models that fit the same data, and tests whether a conclusion survives a change of method, of data or of the policy itself, including whether what a corpus says depends on the technique used to read it.

Publications

Projects