---
id: daal4py
version: "2024.7.0"
license: Apache v2.0
license_treatment: permissive
maintenance: active
---

# daal4py — daal4py is a Convenient Python API to the Intel® oneAPI Data Analytics Library (oneDAL)

License: permissive · Maintenance: active · Downloads: 132.2K/mo

## What it is and what it does

daal4py wraps Intel's oneAPI Data Analytics Library (oneDAL) to provide optimized implementations of machine learning algorithms accessible through a Python API. It targets data scientists and ML practitioners who want to accelerate computationally intensive operations like SVM, linear models, and K-means clustering without rewriting their code. The package depends on daal (the underlying C++ library) and numpy, and is distributed as pre-built wheels for Python 3.9–3.12 on Linux (manylinux1) and Windows (x86_64).

Historically, daal4py offered scikit-learn patching to transparently accelerate scikit-learn calls, but that functionality has been deprecated and moved to a separate package (Intel Extension for Scikit-learn). Current users should treat daal4py as a direct API to oneDAL algorithms rather than a drop-in scikit-learn accelerator.

Use it for:

- Accelerate SVM training and inference on large datasets where CPU optimization is a bottleneck.
- Speed up linear regression, logistic regression, and ridge regression on high-dimensional data.
- Optimize K-means clustering performance for exploratory data analysis on large point clouds.
- Integrate oneDAL algorithms directly into custom ML pipelines that don't rely on scikit-learn.
- Benchmark performance of Intel-optimized algorithms against standard implementations.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

daal4py provides a Python API to Intel's oneAPI Data Analytics Library, offering optimized implementations of machine learning and data analytics algorithms.

Yes, if you need accelerated implementations of specific algorithms (SVM, linear models, K-means) and run on Linux or Windows x86_64. Medium install friction and a stable but slowly-evolving codebase are acceptable tradeoffs for the performance gains. Not recommended if you rely on scikit-learn patching—use Intel Extension for Scikit-learn instead. No known vulnerabilities.

## Install

pip install daal4py
uv add daal4py
poetry add daal4py

## Installing daal4py

Before you install: Medium install friction due to platform-specific wheels (manylinux1 and Windows x86_64 only). Active maintenance with recent commits; last release 696 days ago suggests a stable but not rapidly evolving codebase.

License in practice: Apache v2.0 permissive license allows commercial and private use with minimal restrictions; suitable for most projects.

Quickstart:

pip install daal4py

import daal4py as d4p
from daal4py.sklearn.svm import SVC

# Use optimized SVM classifier
clf = SVC(kernel='rbf')
clf.fit(X_train, y_train)

Requires Linux (manylinux1) or Windows x86_64; macOS and other architectures not supported via PyPI wheels. Requires Python >=3.7.

Verify before relying:

- Whether GPU acceleration is available in current PyPI wheels or only through conda channels
- Performance gains over standard scikit-learn for specific algorithm classes
- Current status of scikit-learn patching after deprecation and move to separate package

## Package facts

- License: Apache v2.0 (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 132.2K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

intel oneapi data analytics library python, accelerated machine learning algorithms, optimized scikit-learn compatible algorithms, high-performance data science library, intel daal python wrapper, performance-optimization, intel-oneapi, machine-learning

[View on SkillFed](https://skillfed.io/packages/daal4py) · [View on PyPI](https://pypi.org/project/daal4py/)