Teradata ModelOps Python SDK

7.2.7 verified Fri May 01 auth: no python

Python client for Teradata ModelOps (TMO), enabling management of machine learning models and experiments on the Teradata Vantage platform. Current version 7.2.7, requires Python >=3.10. Release cadence is roughly monthly.

pip install teradatamodelops

Common errors

error ModuleNotFoundError: No module named 'teradatamodelops' ↓

cause Package not installed or installed in wrong environment.

fix

Run 'pip install teradatamodelops' in the correct Python environment (requires Python >=3.10).

error AttributeError: module 'teradatamodelops' has no attribute 'TMOModel' ↓

cause Using old import pattern or outdated version (<7.0.0).

fix

Upgrade to latest version with 'pip install --upgrade teradatamodelops' and use 'from teradatamodelops import TMOModel'.

error teradataml.exceptions.TeradataMlException: [Teradata ML] Connection context not found ↓

cause Missing call to create_context() before using TMO.

fix

Call 'create_context(host='...', username='...', password='...')' before using teradatamodelops.

Warnings

breaking In version 7.x, the import path changed from 'teradatamodelops' (no subpackage) to the same but some classes were renamed. For example, 'Model' is now 'TMOModel'. ↓

fix Use 'from teradatamodelops import TMOModel' instead of 'from teradatamodelops import Model'.

deprecated The method 'TMOModel.deploy()' is deprecated in 7.2.0+. Use 'TMOModel.deploy_model()' instead. ↓

fix Replace model.deploy(...) with model.deploy_model(...).

gotcha The library requires an active Teradata Vantage connection via teradataml. Calling any TMO method without a valid context will raise a connection error. ↓

fix Ensure you call teradataml.create_context() before using any teradatamodelops class.

Imports

TMOModel

wrong

import teradatamodelops.TMOModel

correct

from teradatamodelops import TMOModel

Module is a Python package; use direct import.

TMOExperiment

from teradatamodelops import TMOExperiment

Quickstart

Basic workflow: connect to Teradata, load data, create a ModelOps model, train, and save.

from teradatamodelops import TMOModel
from teradataml import DataFrame, create_context
# Connect (adjust parameters for your environment)
create_context(host='your_host', username='your_user', password='your_pass')
# Load data
df = DataFrame('your_table')
# Create and train model (example classification)
model = TMOModel(model_type='classification',
                 target_column='target',
                 feature_columns=['feat1','feat2'])
model.fit(df)
# Save model to ModelOps
model_id = model.save(model_name='my_model', project_name='my_project')
print(f'Model saved with ID: {model_id}')