ImportError: Missing optional dependency 'pyarrow'

ImportError: Missing optional dependency ‘pyarrow’

$ python - <<'PY'
import pandas as pd
pd.read_parquet('file.parquet')
PY
Traceback (most recent call last):
  File "<string>", line 2, in <module>
ImportError: Missing optional dependency 'pyarrow'

Why this happens

Pandas delegates reading/writing certain file formats (like Parquet) to optional libraries such as pyarrow or fastparquet. If those packages are not installed, the pandas I/O helper will raise an ImportError asking you to install the required dependency.

Fix

• Install the required package: python -m pip install pyarrow or python -m pip install fastparquet.
• Or explicitly select an engine: pd.read_parquet('file.parquet', engine='pyarrow') (after installing it).

Wrong code

import pandas as pd
# pyarrow not installed
pd.read_parquet('data.parquet')  # raises ImportError

Fixed code

# Install required engine
python -m pip install pyarrow

# After installing, use read_parquet as intended
import pandas as pd
pd.read_parquet('data.parquet')

Notes:

• On some environments (like conda), prefer conda install -c conda-forge pyarrow to binary-compatible wheels.
• If you need to reduce dependencies, use CSV/JSON instead of Parquet where practical.