Overview

PyArrow 20.0, released on April 27, 2025, resolves 259 issues and improves Parquet reading performance.

Main Features

Improved Parquet reading

The Parquet reader is optimized with better parallel decoding and improved memory management for large files.

python 
import pyarrow.parquet as pq

# Optimized Parquet file reading
table = pq.read_table(
    'data.parquet',
    columns=['id', 'value'],
    use_threads=True,
)
print(f'Rows: {table.num_rows}')

Extended compute functions

The pyarrow.compute module gains new transformation and aggregation functions on Arrow arrays.

python 
import pyarrow as pa
import pyarrow.compute as pc

arr = pa.array([3, 1, 4, 1, 5, 9])
sorted_arr = pc.sort_indices(arr)
print(pc.take(arr, sorted_arr))  # [1, 1, 3, 4, 5, 9]
print(f'Mean: {pc.mean(arr).as_py():.2f}')

Sources