MLtraq

🚀 Dive into the world of AI experimentation with MLtraq! This open-source Python library empowers developers to design, execute, and collaborate on experiments seamlessly. Track your progress and share insights effortlessly. 🤖💡 #ML #AI #Python #DataScience #MLtraq

**MLtraq** is an open-source Python library for AI developers that enables designing, executing, and sharing experiments while providing extensive tracking and collaboration features.
It allows for seamless interaction with experiments using Python, Pandas, and SQL without any vendor lock-in, making it flexible and open.
**Key features of MLtraq** include immediate experiment design and execution, collaborative capabilities for backup and sharing, and interoperability with Python, Pandas, and SQL.
The **design choices** in MLtraq involve chained execution of steps using joblib.Parallel and support for custom backends like Dask, Ray, and Spark.
**Persistence in MLtraq** is handled by default with SQLite, but connections to other SQL databases supported by SQLAlchemy are possible.
**Requirements for using MLtraq** include Python 3.10+, SQLAlchemy 2.0+, Pandas 1.5.3+, and Joblib 1.3.2+.
**Installation of MLtraq** can be done via pip install mltraq --upgrade, with a recommendation to pin the exact version for project stability.
**Example 1** demonstrates defining, executing, and querying an experiment with SQL in MLtraq.
**Example 2** showcases parameter grids, parallel execution, and resumed execution in MLtraq experiments.
**Example 3** involves a workflow for IRIS Flowers Classification using MLtraq with various classifiers and evaluation metrics.
The **license** for MLtraq is the BSD 3-Clause License.