Overview

Pandas 1.3, released on July 3, 2021, introduces PyArrow string support and improves method chaining for more fluent code.

Main Features

PyArrow strings

The new ArrowDtype backend for text columns allows using Apache Arrow strings, providing reduced memory consumption and faster string operations.

python 
import pandas as pd

# Using the PyArrow backend for strings
df = pd.DataFrame({
    'name': pd.array(['Alice', 'Bob', 'Charlie'], dtype='string[pyarrow]'),
    'city': pd.array(['Paris', 'Lyon', 'Marseille'], dtype='string[pyarrow]'),
})
print(df.dtypes)
# name    string[pyarrow]
# city    string[pyarrow]

# Optimized string operations
print(df['name'].str.upper())

Method chaining

Pandas 1.3 improves chaining with DataFrame.pipe() and new default inplace=False on more methods, making chainable transformations easier.

python 
import pandas as pd

df = pd.DataFrame({
    'product': ['A', 'B', 'C', 'A', 'B'],
    'sales': [100, 200, 150, 300, 250],
})

# Fluent transformation chaining
result = (
    df
    .query('sales > 100')
    .groupby('product')['sales']
    .sum()
    .reset_index()
    .rename(columns={'sales': 'total'})
)
print(result)

Sources