Skip to content

Simplify and accelerate your machine learning development with mleko. Designed with modularity and customization in mind, it seamlessly integrates into your existing workflows. Its robust caching system optimizes performance, taking you from data ingestion to finalized models with unparalleled efficiency.

License

Notifications You must be signed in to change notification settings

klarna-incubator/mleko

mleko: Streamlining Machine Learning Pipelines in Python

Simplify and accelerate your machine learning development with mleko. Designed with modularity and customization in mind, it seamlessly integrates into your existing workflows. Its robust caching system optimizes performance, taking you from data ingestion to finalized models with unparalleled efficiency.

Developed at Klarna License Static Badge

Latest Status PyPI - Downloads Python Version

Tests pre-commit Black

Features

mleko is engineered to address the end-to-end needs of machine learning pipelines, providing robust, scalable solutions for data science challenges:

  • Ingest: Seamlessly integrates with data sources like AWS S3 and Kaggle, offering hassle-free data ingestion and compatibility.
  • Export: Supports exporting data to various formats and platforms, locally or in the cloud, to ensure that your data is accessible and shareable.
  • Convert: Specializes in data format transformations, prominently featuring high-performance conversions from CSV to Vaex DataFrame, to make your data pipeline-ready.
  • Split: Employs sophisticated data partitioning algorithms, allowing you to segment DataFrames into train, test, and validation sets for effective model training and evaluation. -Filter: Provides a suite of filtering techniques such as resampling or simple expression-based filtering, enabling you to focus on the most relevant data.
  • Feature Selection: Equipped with a suite of feature selection techniques, mleko enables model performance by focusing on the most impactful variables.
  • Transformation: Facilitates data manipulations such as Frequency Encoding and Standardization, ensuring that your data conforms to the prerequisites of the machine learning algorithms.
  • Model: Provides a core set of functionalities for machine learning models, including in-built support for hyperparameter tuning, thereby streamlining the path from data to deployable model.
  • Pipeline: Unifies the entire workflow into an intuitive directed acyclic graph (DAG) architecture, promoting reproducibility and reducing iteration time and time-to-market for machine learning models.

By integrating these features, mleko serves as a comprehensive toolkit for machine learning practitioners looking to build robust models efficiently.

Installation

You can install mleko via pip from PyPI:

$ pip install mleko

Usage & Examples

See the documentation for more information or check out the usage examples on well-known datasets like the Titanic Dataset.

Issues

If you encounter any problems, please file an issue along with a detailed description.

Contributing

We are open to, and grateful for, any contributions made by the community. To learn more, see the Contributor Guide.

Release History

See our changelog.

Acknowledgements

The development of mleko was influenced by existing work of the following individuals:

Their insights and contributions provided a solid foundation for this library. We appreciate their effort and recognize their contributions that led to the creation of mleko.

License

Copyright © 2024 Klarna Bank AB

For license details, see the LICENSE file in the root of this project.

About

Simplify and accelerate your machine learning development with mleko. Designed with modularity and customization in mind, it seamlessly integrates into your existing workflows. Its robust caching system optimizes performance, taking you from data ingestion to finalized models with unparalleled efficiency.

Topics

Resources

License

Code of conduct

Security policy

Stars

Watchers

Forks

Contributors 4

  •  
  •  
  •  
  •