Guest Lecture: Advanced Applied Econometrics, Master of Agricultural and Food Economics, SS2024, Uni Bonn
Hugo Storm (hugo.storm@ilr.uni-bonn.de), May 2024
- Understand intuition of Regularized estimators (LASSO, ridge regression, etc)
- Understand how LASSO can be used for model selection
- Understand difference between model selection for prediction and causal analysis
- Understand connection to econometrics model selection
- Understand “Post-Lasso” model selection and its limitations
- Understand better approach for model selection using Lasso when doing causal analysis
available under: https://docs.google.com/presentation/d/15KhGbBiptpBqisdCcfm1GV7wqHx7bh2P9RuMSQ1-hxw/edit?usp=sharing
See folder: /examples
housing_lasso.Rmd(Example on how to use Lasso for variable selection in an prediction context)wages_lasso.Rmd(Example on how to use Lasso for variable selection in causal interpretation context)
-
Note 1: the datasets to run the notebooks are available on request.*
-
Note 2: in the folder
/examplesthere are also html files for each notebook if you clone this repository (or download it) you can open those files in a browser. Those files give the code as well as rendered outputs. (Sadly github has in issue with those files an no preview is provided, so you actually need to download those files). *
available under: https://youtu.be/oDVATVVYOnk
Paper on double selection Belloni, Alexandre, Victor Chernozhukov, and Christian Hansen. 2014. “High-Dimensional Methods and Inference on Structural and Treatment Effects.” The Journal of Economic Perspectives 28 (2): 29–50. https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.28.2.29
Full book on LASSO Hastie, Trevor, Robert Tibshirani, and Martin Wainwright. 2015. Statistical Learning with Sparsity: The Lasso and Generalizations. Chapman & Hall/CRC Monographs on Statistics and Applied Probability. Philadelphia, PA: Chapman & Hall/CRC. pdf available at http://web.stanford.edu/~hastie/StatLearnSparsity/