I noticed that there is a good amount of code pointing to a read_parquet implementation available with the Pandas API, but I am seeming to have some trouble with it. Is this something that is supported or is the package limited to read_csv as mentioned in the documentation?

I first received the error stating that I must have pyarrow or fastparquet to run the read_parquet function, so I decided on fastparquet since the installation instructions already put pyarrow into the intel sdc conda environment.

Now as I attempt to run with nopython, I am met with this compiler error:

unknown attribute 'read_parquet' of type module(<module 'pandas' from '/home/jds35172/anaconda3/envs/intel-sdc-env/lib/python3.7/site-packages/pandas/init.py'>)

Do I have to use the Makefiles to integrate this into the environment, or are there any alternatives to simply resorting to nopython?

Thank you!