mltable

What it is and what it does

MLTable is a Python data loading toolkit designed to simplify access to tabular data by abstracting schema definition and providing a unified interface across multiple file formats. It loads CSV, parquet, delta lake, and JSON-lines files from local disk or cloud storage, then materializes them into Pandas DataFrames with minimal boilerplate. The package is built on top of Azure SDK components (azureml-dataprep, azure-core, azure-mgmt-core) and includes utilities for sampling, filtering, and schema capture on large datasets.

The package is actively maintained and supports Python 3.10 through 3.13. It carries a proprietary Microsoft license with unclear treatment outside Azure ML contexts. With low install friction and no known security vulnerabilities, it is production-stable and intended for developers and researchers working with tabular data in machine learning pipelines.

Use it for:

• Load CSV or parquet files into Pandas for exploratory data analysis without manual schema definition
• Sample or filter large datasets before materializing them into memory for model training
• Unify data loading logic across multiple file formats (CSV, parquet, delta lake, JSON-lines) in ML workflows
• Integrate tabular data loading into Azure ML training pipelines with native schema support
• Define and persist data schemas in MLTable YAML format for reproducible data ingestion

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

MLTable provides fast, flexible data loading and schema abstraction for tabular data from CSV, parquet, delta lake, and JSON-lines files, materializing them into Pandas DataFrames.

Yes, if you are working within Azure ML or need unified tabular data loading across multiple formats. The low install friction, active maintenance, and lack of known vulnerabilities make it safe to use. However, verify the proprietary license terms for your use case—it is designed for Azure ML contexts and may have restrictions outside that ecosystem. Not necessary if you only work with a single file format or prefer standard Pandas I/O.

Install

mltable on PyPI

pip install mltable

uv add mltable

poetry add mltable

Installing mltable

Before you install

Low install friction with a pure Python wheel. Active maintenance with a release 7 days ago. Depends on 10 runtime packages including Azure SDK components, cryptography, and standard utilities; all are widely used and stable.

License in practice

Licensed under a proprietary Microsoft Azure ML license (unclear treatment). Review the full license at https://aka.ms/azureml-preview-sdk-license before use in commercial or redistributed contexts.

Quickstart

pip install mltable

from mltable import MLTable

table = MLTable.from_delimited_files(paths=['data.csv'])
df = table.to_pandas()

Requires Python 3.10 or later (supports 3.10, 3.11, 3.12, 3.13). Pre-installed on AzureML compute instances.

Verify before relying

• Whether the proprietary license permits use outside Azure ML environments or in closed-source applications
• Performance characteristics and scalability limits for very large files or datasets
• Whether delta lake support requires additional dependencies or cloud storage configuration

Package facts

Evidence: mltable-1.7.0-py3-none-any.whl

Tags

More Scientific/Engineering packages