Skip to main content
PyArrow is the Python library for Apache Arrow, providing a Python API for Arrow’s functionality along with tools for integration with pandas, NumPy, and other Python data tools.

​
Quick Install

Install PyArrow using pip or conda: 
pip install pyarrow
These commands install the latest stable version of PyArrow with pre-compiled binary wheels for your platform.

​
Supported Platforms

PyArrow provides pre-built binary wheels for:
  • Linux: x86_64, aarch64 (ARM64)
  • macOS: Intel (x86_64) and Apple Silicon (ARM64)
  • Windows: x86_64
Supported Python versions: 3.9, 3.10, 3.11, 3.12, and 3.13

​
Installation Methods

​
Using pip (PyPI)

Install from the Python Package Index (PyPI):
1

Install PyArrow

pip install pyarrow
This installs the latest stable version.
2

Install a specific version

# Install a specific version
pip install pyarrow==15.0.0

# Install the latest version in a major series
pip install "pyarrow>=15.0.0,<16.0.0"
3

Verify installation

python -c "import pyarrow as pa; print(pa.__version__)"
Windows Users: If you encounter import errors, you may need to install the Visual C++ Redistributable for Visual Studio.

​
Using conda (conda-forge)

Install from the conda-forge channel:
1

Install from conda-forge

conda install pyarrow -c conda-forge
The -c conda-forge flag ensures you get the latest version from conda-forge.
2

Create a new environment with PyArrow

conda create -n arrow-env python=3.11 pyarrow -c conda-forge
conda activate arrow-env
3

Verify installation

python -c "import pyarrow as pa; print(pa.__version__)"
conda-forge provides binary packages across all supported platforms and often includes additional features like S3 support.

​
Installing Optional Dependencies

PyArrow has optional dependencies for additional functionality: 
# Install PyArrow with pandas
pip install pyarrow pandas

​
Testing Your Installation

Verify PyArrow is working correctly:
1

Import PyArrow

import pyarrow as pa
print(f"PyArrow version: {pa.__version__}")
2

Create a simple array

import pyarrow as pa

# Create an Arrow array
arr = pa.array([1, 2, 3, 4, 5])
print(arr)
print(f"Type: {arr.type}")
3

Test Parquet support

import pyarrow as pa
import pyarrow.parquet as pq

# Create a simple table
table = pa.table({'a': [1, 2, 3], 'b': ['x', 'y', 'z']})
print("Parquet support available!")
4

Test pandas integration

import pyarrow as pa
import pandas as pd

# Create a pandas DataFrame
df = pd.DataFrame({'a': [1, 2, 3], 'b': [4, 5, 6]})

# Convert to Arrow table
table = pa.Table.from_pandas(df)
print(f"Arrow table shape: {table.shape}")

# Convert back to pandas
df2 = table.to_pandas()
print("Pandas integration working!")

​
Common Use Cases

​
Reading and Writing Parquet Files

import pyarrow as pa
import pyarrow.parquet as pq
import pandas as pd

# Create a DataFrame
df = pd.DataFrame({
    'name': ['Alice', 'Bob', 'Charlie'],
    'age': [25, 30, 35],
    'city': ['New York', 'London', 'Paris']
})

# Write to Parquet
table = pa.Table.from_pandas(df)
pq.write_table(table, 'data.parquet')

# Read from Parquet
table2 = pq.read_table('data.parquet')
df2 = table2.to_pandas()
print(df2)

​
Working with CSV Files

import pyarrow.csv as csv
import pyarrow as pa

# Read CSV file
table = csv.read_csv('data.csv')

# Convert to pandas if needed
df = table.to_pandas()

​
Memory-Mapped Files for Large Datasets

import pyarrow as pa
import pyarrow.parquet as pq

# Read large Parquet file without loading into memory
parquet_file = pq.ParquetFile('large_data.parquet')

# Read in batches
for batch in parquet_file.iter_batches(batch_size=10000):
    # Process batch
    df = batch.to_pandas()
    # Your processing here

​
Version Compatibility

PyArrow follows semantic versioning. The Arrow IPC format is stable, but API changes may occur between major versions. Always check the changelog when upgrading.

​
Python Version Support

  • Python 3.9+: Recommended for the latest PyArrow versions
  • Python 3.8: Supported in PyArrow < 15.0
  • Python 3.7 and earlier: No longer supported

​
Troubleshooting

If you get an import error on Windows, install the Visual C++ Redistributable:
  1. Download from Microsoft’s official page
  2. Install the x64 version
  3. Restart your Python environment
Ensure PyArrow is installed in the correct Python environment:
# Check which Python you're using
which python

# Check if PyArrow is installed
pip list | grep pyarrow

# Reinstall if needed
pip install --force-reinstall pyarrow
If you have conflicts with NumPy or pandas:
# Update all related packages
pip install --upgrade pyarrow pandas numpy

# Or create a fresh environment
conda create -n fresh-env python=3.11 pyarrow pandas numpy
conda activate fresh-env
First import of PyArrow may be slow. This is normal. Subsequent imports are faster:
import pyarrow as pa  # First import may take a few seconds
# Subsequent operations are fast

​
Building from Source

For development or custom builds:
Building PyArrow from source requires Arrow C++ to be built first. This is an advanced option and pre-built wheels are recommended for most users.
# Clone the repository
git clone https://github.com/apache/arrow.git
cd arrow/python

# Install build dependencies
pip install -r requirements-build.txt

# Build and install
python setup.py build_ext --build-type=release install
See the Python Development Guide for detailed instructions.

​
Next Steps

Now that you have PyArrow installed:

Build docs developers (and LLMs) love

Get started for freeTalk to us