machine-learning and Bayesian frameworks

Through my varied exposure to Economics, and Econometrics, I have become intrigued by the recent traction gained by causal inference literature as it becomes more integrated with statistical methods. My research interests lie in developing machine-learning and Bayesian frameworks to address causal inference problems, particularly in Development Economics and Public Economics.

My present research project direction would be to investigate the use of Bayesian machine-learning algorithms for estimating and minimising the loss for policy assignment when using observational instead of experimental data. Inspired by the literature on dynamic policy assignment, such bandit algorithms, I seek to understand how to better aggregate observational data to minimise the loss for policy decisions when experimental data is unavailable. Another interesting, very related avenue, would be to understand how to better aggregate experimental and observational data for this same purpose. To do so, I am interested in the use of Bayesian hierarchical models, and considering that several of these problems are high-dimensional, I intend to combine these