Mixed Data Sampling (MIDAS) Modeling in Python

Libirary usage tutorial:

Current Features

• Beta, Exponential Almon, and Hyperbolic scheme polynomial weighting methods.
• Lagged matrix generator function to project higher frequency data onto lower frequency.
• Flexible MIDAS ordinary least squares regressor model and results wrapper classes.
• Basic statisical summary methods like R^2 score and variable significance (t-test p-values).

Future Work

• Improve efficiency of exogenous lagged projection generator function.
• Enable horizon to be set for each exogenous variable separately.
• Enable prelagged variables to be used for faster model fitting.
• More comprehensive statistical summary method simialr to statsmodels.api.OSL().fit().summary().
• Create a Flexible MIDAS logistic classifier model class.