- The repository contains implementaions of Mixed-data sampling (MIDAS).12
- Currently, it is based on the package by Jonas Striaukas (https://github.com/jstriaukas/midasml). 34 The language of that package is R. Although it has Python Version, it contains a lot of bugs. In addition, some of the code is also not fully optimized.
- The main goal of this repository is to build a python version MIDAS package
- I'll start with debugging the code mentioned above, and try to optimize it.
- I also have the plan to introduce PyTorch to this repository, which would benefit building a hybrid model combines DNNs and MIDAS together.
- fixed bugs in date_functions.mixed_freq_data
- optimized date_functions.is_na & date_functions.date_vec
- fixed bugs in midas_polynomials.gb
Footnotes
-
Eric Ghysels, Pedro Santa-Clara, Rossen Valkanov. The MIDAS touch: Mixed data sampling regression models.(2004) ↩
-
Eric Ghysels, Arthur Sinko, Rossen Valkanov. MIDAS regressions: Further results and new directions.(2007) ↩
-
Babii, A., Ghysels, E., & Striaukas, J. Machine learning time series regressions with an application to nowcasting, (2022) Journal of Business & Economic Statistics, Volume 40, Issue 3, 1094-1106. https://doi.org/10.1080/07350015.2021.1899933. ↩
-
Babii, A., Ghysels, E., & Striaukas, J. High-dimensional Granger causality tests with an application to VIX and news, (2022) Journal of Financial Econometrics, Forthcoming. ↩