Learning from high-dimensional, heterogeneous data: Machine learning methods in econometrics

• Applicants: Professor Dr. Helmut Farbmacher; Professor Dr. Martin Spindler
• Subject Area: Statistics and Econometrics
• Term: from 2020 to 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 431701914

• Final Report Year: 2025

Final Report Abstract

In this project, in particular in the first part, progress on the development of methods for learning from high-dimensional complex data was achieved. Modern machine learning methods have been developed, their theoretical properties analyzed and the methods have been implemented in open source software. Additionally, empirical applications have been considered to show the advantages of the methods. The project led to many publications in internationally renowned journals.

Publications

• Instrument Validity Tests With Causal Forests. Journal of Business & Economic Statistics, 40(2), 605-614.
  Farbmacher, Helmut; Guber, Raphael & Klaassen, Sven
  
• Distributed Double Machine Learning with a Serverless Architecture. Companion of the ACM/SPEC International Conference on Performance Engineering, 27-33. ACM.
  Kurz, Malte S.
  
• Heterogeneous effects of poverty on attention. Labour Economics, 71, 102028.
  Farbmacher, Helmut; Kögel, Heinrich & Spindler, Martin
  
• An explainable attention network for fraud detection in claims management. Journal of Econometrics, 228(2), 244-258.
  Farbmacher, Helmut; Löw, Leander & Spindler, Martin
  
• Causal mediation analysis with double machine learning. The Econometrics Journal, 25(2), 277-300.
  Farbmacher, Helmut; Huber, Martin; Lafférs, Lukáš; Langen, Henrika & Spindler, Martin
  
• DoubleML – Double Machine Learning in Python. Python package, version 0.5.0,
  Bach, P., V. Chernozhukov, M. S. Kurz & M. Spindler
  
• DoubleML – Double Machine Learning in R. R package, version 0.5.1
  Bach, P., V. Chernozhukov, M. S. Kurz & M. Spindler
  
• DoubleML-Serverless – Distributed Double Machine Learning with a Serverless Architecture. Python package, version 0.0.3
  Kurz, M. S.
  
• pacotest: Testing for Partial Copulas and the Simplifying Assumption in Vine Copulas. R package, version 0.5.0
  Kurz, M. S.
  
• Testing the simplifying assumption in high-dimensional vine copulas. Electronic Journal of Statistics, 16(2).
  Kurz, Malte S. & Spanhel, Fabian
  
• Using Double Machine Learning to Understand Nonresponse in the Recruitment of a Mixed-Mode Online Panel. Social Science Computer Review, 41(2), 461-481.
  Felderer, Barbara; Kueck, Jannis & Spindler, Martin
  
• Vine copula based knockoff generation for high-dimensional controlled variable selection
  Kurz, M. S.
  
• vineknockoffs: Vine copula based knockoffs. Python package, version 0.1.0
  Kurz, M. S.
  
• “DoubleML – An Object-Oriented Implementation of Double Machine Learning in Python”. In: Journal of Machine Learning Research 23.53, pp. 1– 6
  Bach, P., V. Chernozhukov, M. S. Kurz & M. Spindler
  
• Estimation and inference of treatment effects with L2-boosting in high-dimensional settings. Journal of Econometrics, 234(2), 714-731.
  Kueck, Jannis; Luo, Ye; Spindler, Martin & Wang, Zigan