This is the official code repository for paper titled "Regularized estimation of high-dimensional factor-augmented vector autoregressive (FAVAR) models" (2020, Journal of Machine Learning Research), by Jiahe Lin and George Michailidis.

• Link to paper: https://jmlr.csail.mit.edu/papers/volume21/19-874/19-874.pdf
• To cite this work: Lin, J., Michailidis, G. (2020) Regularized estimation of high-dimensional factor-augmented vector autoregressive (FAVAR) models. Journal of Machine Learning Research, 21(117): 1–51.

  @article{lin2020regularized,
      title={Regularized estimation of high-dimensional factor-augmented vector autoregressive {(FAVAR)} models},
      author={Lin, Jiahe and Michailidis, George},
      journal={The Journal of Machine Learning Research},
      volume={21},
      number={1},
      pages={4635--4685},
      year={2020},
      publisher={JMLRORG}
  }
  

In this repository, we provide both Python and R implementation of the proposed two-stage methodology.

• For Python version, see demo.ipynb, whose source files are under ./srcPy
• For R version, see example.R, whose source files are under ./srcR

Note that there is some difference in the behavior of Lasso from sklearn in Python and glmnet from R, due to the discrepancy in their respective underlying implementations. The original paper, when it was being developed, replied on the R version.

• For questions on the implementation, contact Jiahe Lin [jiahelin AT umich DOT edu]