Conferences

This talk explores how to estimate large two-way fixed effects models efficiently in Python using pyfixest. Using airline pricing as an applied setting, it explains why conventional dummy-variable approaches become computationally expensive and how fixed-effect absorption avoids constructing very large design matrices.

The presentation covers within transformations, alternating projections, and the extension from linear regression to Poisson models through iteratively reweighted least squares. It also discusses practical performance considerations and lessons from applying these methods to large airline datasets.

The applied examples draw on Decision Science work at ADC Consulting, where these methods are used to support pricing and experimentation decisions in practice.

How do customers respond to price changes, and how can insurers measure those effects reliably? This session explores how price elasticity curves help insurers understand the relationship between premiums, demand, and customer value.

The presentation discusses the practical interpretation of elasticity, the role of data and experimentation, and the challenges of estimating price effects in a market influenced by customer behaviour, competition, and risk selection.

The session draws on Decision Science work at ADC Consulting and connects statistical methodology with practical pricing decisions in insurance.