Indeed, I was trying to try there TimeGPT. How do I get the token?

The Nixtla statsforcast package also ran into some issues:

---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
/tmp/ipykernel_6602/831170903.py in <module>
      1 import pandas as pd
      2 
----> 3 Y_df = pd.read_parquet('<https://datasets-nixtla.s3.amazonaws.com/m4-hourly.parquet>')
      4 
      5 Y_df.head()

~/anaconda3/lib/python3.9/site-packages/pandas/io/parquet.py in read_parquet(path, engine, columns, storage_options, use_nullable_dtypes, **kwargs)
    491         **kwargs,
    492     )
--> 493 
    494     if path is None:
    495         assert isinstance(path_or_buf, io.BytesIO)

~/anaconda3/lib/python3.9/site-packages/pandas/io/parquet.py in get_engine(engine)
     51     """return our implementation"""
     52     if engine == "auto":
---> 53         engine = get_option("io.parquet.engine")
     54 
     55     if engine == "auto":

ImportError: Unable to find a usable engine; tried using: 'pyarrow', 'fastparquet'.
A suitable version of pyarrow or fastparquet is required for parquet support.
Trying to import the above resulted in these errors:
 - No module named 'pandas.core.arrays._arrow_utils'
 - Missing optional dependency 'fastparquet'. fastparquet is required for parquet support. Use pip or conda to install fastparquet.

I am testing on an amazon webspace. Not sure if that has any envirobment issues. Thanks.

It appears many people had the same issue. Sent an email to request a token ...