My research
I work on causal machine learning, especially how to use multimodal data (text, images, tables) to get better causal estimates. Most of my work involves Double Machine Learning.
Working papers
Current research projects, mostly around causal inference with modern ML methods.
Adventures in Demand Analysis Using AI
Philipp Bach, Victor Chernozhukov, Sven Klaassen, Martin Spindler, Jan Teichert-Kluge, Suhas Vijaykumar
We use text, images, and tabular data to model product demand. The idea is to get better price elasticity estimates by using all available information about products, not just the numbers.
DoubleMLDeep: Estimation of Causal Effects with Multimodal Data
Sven Klaassen, Jan Teichert-Kluge, Philipp Bach, Victor Chernozhukov, Martin Spindler, Suhas Vijaykumar
A neural network setup designed for DoubleML. We also created a semi-synthetic benchmark for testing how well methods handle text and image confounders.
Talks
Presentations at conferences and universities.
DoubleMLDeep: Estimation of Causal Effects with Multimodal Data
Talked about how to use text and images in DoubleML for treatment effect estimation. Presented the new datasets and tools we built.
Econometric Society ESIF AI/ML, Cornell University, 2024
Estimating Price Elasticities with Text and Images
How to extend DoubleML for multimodal confounding, with applications to e-commerce pricing.
University of Queensland, Econometrics Colloquium, 2024
Estimating Price Elasticities with Text and Images
A hands-on talk about price elasticity estimation with multimodal data using DoubleML.
Australian National University, Computational Economics Workshop, 2024
Price Elasticity Estimation using Image and Text Data
Early version of our multimodal causal inference work.
Causal Data Science Meeting, 2023
Other publications
Causal Machine Learning with Deep Learning approaches using the DoubleML Framework
Moritz Sundermann, Jan Teichert-Kluge
Master’s thesis where we combined deep learning with DoubleML. Includes simulations and a real-world case study.