teradataml

What it is and what it does

teradataml is a Python interface to Teradata Vantage's in-database analytics engine. It lets you perform machine learning, data transformation, and statistical analysis directly on Teradata without writing SQL—operations execute server-side on Vantage and results return to Python. The library includes AutoML functions (clustering, classification, regression, fraud detection, churn prediction), a feature store for managing ML features, data manipulation via DataFrame-like objects, and integration with pandas and other open-source libraries.

The package depends on teradatasql and teradatasqlalchemy for database connectivity, plus standard Python libraries like SQLAlchemy, pandas, requests, and cryptography. It supports OAuth 2.0 authentication (including Device Code Grant for PING and KEYCLOAK IDPs) and optional extras for visualization and EDA. Recent releases added array operations, cross-validation support in hyperparameter tuning, and FilterManager for feature processing workflows.

Use it for:

• Build and deploy machine learning models (classification, regression, clustering) directly on Teradata without exporting data.
• Perform feature engineering and feature store management for ML pipelines using Teradata as the compute backend.
• Run AutoML workflows to automatically select algorithms, tune hyperparameters, and generate SHAP explanations on large datasets.
• Transform and filter data at scale using DataFrame operations and array functions without SQL coding.
• Integrate Teradata analytics into Python data science workflows alongside pandas and scikit-learn.

Worth the install?

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

teradataml provides Python access to analytic functions running on Teradata Vantage, enabling data analysis, machine learning, and data transformation without writing SQL.

Yes, if you have a Teradata Vantage instance and need Python-native analytics without SQL. The package is actively maintained, has low install friction, and offers a broad feature set (AutoML, feature store, data manipulation). However, verify the Teradata License Agreement terms for your use case first, and confirm that optional dependencies (scikit-learn, lightgbm) are available if you need AutoML or visualization features.

Install

teradataml on PyPI

pip install teradataml

uv add teradataml

poetry add teradataml

Installing teradataml

Before you install

Low install friction with a pure-wheel distribution. Active maintenance with a recent release (63 days old). Depends on 13 runtime packages including SQLAlchemy, pandas, and OAuth libraries, which are stable and widely used.

License in practice

Licensed under Teradata License Agreement with unclear treatment—not an open-source SPDX identifier. Verify licensing terms with Teradata before use in commercial or redistributed contexts.

Quickstart

pip install teradataml

from teradataml import create_context
from teradataml import DataFrame

ctx = create_context(host='vantage_host', username='user', password='pass')
df = DataFrame(table_name='my_table')

Requires a Teradata Vantage instance to connect to; local-only use is not supported. Requires Python 3.8 or later.

Verify before relying

• Whether the unclear license permits use in proprietary or commercial applications without explicit Teradata agreement.
• Whether optional dependencies (scikit-learn, lightgbm, matplotlib, seaborn) are required for core functionality or only for specific features.
• Performance and scalability limits when working with large datasets on Teradata Vantage.

Package facts

Evidence: teradataml-20.0.0.11-py3-none-any.whl

Keywords: Teradata

Tags

More Front-Ends packages