PyArrow Backend

Pandas 2.0 lets you use PyArrow as a storage backend via dtype_backend='pyarrow'. Arrow types provide better missing value handling, more efficient strings, and native interoperability with the Arrow ecosystem.

Usage

python 
import pandas as pd

# Read a CSV with PyArrow backend
df = pd.read_csv(
    'data.csv',
    dtype_backend='pyarrow',
    engine='pyarrow',
)
print(df.dtypes)
# name    string[pyarrow]
# age     int64[pyarrow]
# salary  double[pyarrow]

# Benefits:
# - Native string types (no object dtype)
# - Native missing values (no NaN for integers)
# - Reduced memory on repetitive strings
print(df.memory_usage(deep=True))

Sources