Unexpected Uses of Data Pipelines in Economic Modeling

Data pipeline frameworks, such as Luigi for Python, are an essential tool for data scientists interested in chaining together data- and compute- intensive tasks. In addition to this more common use, these frameworks can be a powerful tool for economists to solve problems in economic modeling, such as Monte Carlo analysis, decomposition of sources of error in complex models using bootstrap methods, and solutions to structural models using iterative methods.

This talk will first review the concepts behind data pipeline frameworks, using Luigi as an example, then discuss the new ways in which the Penn-Wharton Budget Model is using Luigi with Python to solve the previously mentioned problems in the context of its microsimulation model, a reduced-form model of the US population used by PWBM to produce economic projections of US demographics.

About our Presenter: Nick Janetos is a research economist and the head of the microsimulation group at the Penn-Wharton Budget Model, a non-partisan think tank. He received his PhD from Penn in 2017 and specializes in microeconomic theory and computational economics.

Pizza and light refreshments will be served at 6pm. Presentation will start at 6:30pm.