Select Page

Causal inference: the uplift

An R&D programme carried out in partnership with Paris Dauphine University.

Context

‘Uplift modelling’ is the name given to a set of scoring methods used to assess individual sensitivity to sales prospecting.

Why?

The modelling used is based on mathematical formalisation, the aim of which is to maximise the ROI of a sales campaign.

How?

The aim of the research programme is to maximise the effectiveness of a company’s marketing campaigns using artificial intelligence models. The research programme draws on 14 types of artificial intelligence model, from the simplest to the most complex, in order to offer the best compromise between performance and complexity, and to adapt to each company’s existing data.

Who?

Companies in all sectors using marketing campaigns targeting their customers.

RESEARCHER

Clément Feutry (Researcher (PhD) — Consultant)

Clément Feutry (Researcher (PhD) — Consultant)

After gaining experience in academic research and teaching at ENS Cachan, Clément is now one of the driving forces behind Square Management’s R&D activities. He has expertise in image processing and machine learning, the subject of his doctorate. In addition to his work on uplift, he is also involved in other R&D topics at the Square Research Center: explainability. Clément works on these subjects on a contract basis, contributing his expertise in automation and modelling. His work can include POCs or start with an audit of the existing system.

Publications
  • “Après l’explosion du e‑commerce, l’heure du bilan sur la génération de trafic et son ROI a sonné!” LSA, 02/12/2021
  • “Revisiting the CATE/uplift problem with binary outcome and treatment: new metric, models and a comprehensive benchmark” en cours de publication

OTHER SQUARE RESEARCH CENTER PROGRAMS

Share This