python_integration

[harrisonzhu508]

Main features:

• Implemented a prelim data wrangling class using Python, producing a dictionary for ingestion into Stan compiler

• Notebook available at notebooks/Test.ipynb for fitting. It is observed that it is slightly slower than the rstan compiler, so perhaps one can reticulate in R to slot the dataset into the rstan compiler Aim:

• Provide a Python data pipeline for more people to use our code to produce analysis

• Enable more sophisticated statistical ML modelling in Python since most up-to-date ML libraries (e.g. ones having variational inference) are available in Python

Future work:

• No code or results checks were done, so this would be future work
• unit testing
• proper documentation and docstrings

Any comments welcome.

[am5113]

Maybe it's a good idea to keep everything Python self-contained in the "Python" folder so everything isn't mixed together. For example, having a .yml file for R and requirements.txt for Python doesn't seem too elegant. What do you think?