Modelling Big Data using Autometrics

This event is part of the Specialist Training Series on Machine Learning and Data Science delivered by the South & East Network for Social Sciences Doctoral Training Partnership (SENSS).

SENSS Specialist Training: Modelling Big Data using Autometrics
Instructor: Dr Elisabetta Pellini (Bayes Business School)
Term: Autumn 2025

Module Outline and Aims

In the age of big data, researchers and analysts increasingly face challenges of model selection, overfitting, and interpreting complex high-dimensional datasets. This session introduces Autometrics, a powerful econometric modelling tool designed to automate model selection in large datasets while maintaining strong statistical foundations.

Autometrics is part of the OxMetrics suite and builds on the general-to-specific (Gets) modelling philosophy. It enables users to extract robust models from large sets of candidate variables, making it especially suitable for empirical work in economics, finance, and the social sciences where theory may not fully specify the data-generating process.

This practical session will combine core theoretical principles with hands-on demonstrations, using real-world data examples to illustrate how Autometrics addresses the challenges of modelling with many variables.

Prerequisites

Students are expected to have a knowledge of statistics (descriptive and inference) and basic econometrics.

Software

Participants will use OxMetrics during the session. Instructions for installation will be provided in advance.

Content Outline

• General-to-specific (Gets) modelling and the problem of model selection
• Autometrics algorithm: overview and logic
• Diagnostic testing and model evaluation
• Hands-on modelling with real-world big data using Autometrics

Learning Objectives

By the end of the session, participants will be able to:

1. Understand the principles behind general-to-specific model selection and how Autometrics operationalises these.
2. Recognise the challenges of modelling large datasets, including overfitting, multicollinearity, and irrelevant variables.
3. Critically evaluate the strengths and limitations of automated model selection tools in empirical research.

Main References

• Doornik, J.A., 2009a. Autometrics, in: J.L. Castle, N. Shephard (Eds.), The Methodology and Practice of Econometrics, Oxford University Press, Oxford (2009), pp. 88–121.
• Hendry, D.F., and J.A. Doornik (2014). Empirical Model Discovery and Theory Evaluation. MIT Press.
• Hendry, D.F., and J.A. Doornik(2018). Empirical Econometric Modelling – PcGive 16: Volume I. Timberlake Consultants Press, London.

Please ensure you meet the prerequisites before registering.

Sign up form: https://essex.eu.qualtrics.com/jfe/form/SV_9Zit5afuGNbuIfk

Please direct enquiries to: trainingmanager@senss-dtp.ac.uk