Using the Bring Your Own Model User Guide
The Vantage Bring Your Own Model (BYOM) package gives data scientists and analysts the ability to operationalize predictive models in Vantage using predictive models trained in external tools.
- Data Analysts
- Data Scientists
Why Would I Use this Content?
To create or convert your predictive model using a supported model interchange format and store it in a Vantage table.
How Do I Use this Content
Learn the functions to take predictive models or converted predictive models stored in a Vantage table and use BYOM Predict, PMMLPredict, H2OPredict, ONNXPredict, DataikuPredict, or DataRobotPredict to score your data with the model.
How Do I Get Started?
Contact Teradata to order the Vantage BYOM software package and schedule installation.
Bring Your Own Model
The Vantage Bring Your Own Model (BYOM) package gives data scientists and analysts the ability to operationalize predictive models in Vantage. Predictive models trained in external tools with sample data can be used to score data stored in Vantage using the BYOM Predict. Create or convert your predictive model using a supported model interchange format (PMML, MOJO, ONNX, Dataiku, and DataRobot are currently available), store it in a Vantage table, and use the BYOM PMMLPredict, H2OPredict, ONNXPredict, DataikuPredict, or DataRobotPredict to score your data with the model.
- Predictive Model Markup Language (PMML)
- Model Object Optimized (MOJO)
- Open Neural Network Exchange (ONNX)
- Dataiku
- DataRobot
Without Vantage BYOM
- Prepare or transform sample data/features in Vantage.
- Export sample data from Vantage to the external platform.
- Construct and train models using external modeling tools running on an external platform.
- Prepare or transform scoring data/features in Vantage.
- Export the raw data/features from Vantage to the external platform.
- Run scoring/prediction jobs on the external platform.
- Import scores from the external platform to Vantage.
- Repeat steps 4-7 as new raw data/features become available.
Vantage BYOM is more efficient because you can eliminate steps 5-8, and scoring models in Vantage has high-performance and is scalable to multiple models running concurrently.
With Vantage BYOM PMMLPredict Function
- Generate the external model in PMML or convert the external model to PMML.
- Load the PPML model into a Vantage table.
- Use the PMMLPredict function to score/predict on the data in Vantage.
With Vantage BYOM H2OPredict Function
- Generate the external MOJO model in H2O Open Source or H2O Driverless AI.
- Load the MOJO model into a Vantage table.
- Use the H2OPredict function to create predictions on the data stored in Vantage.
With Vantage BYOM ONNXPredict Function
- Generate the external model in ONNX or convert the external model to ONNX.
- Load the ONNX model into a Vantage table.
- Use the ONNXPredict function to create predictions on the data stored in Vantage.
With Vantage BYOM DataikuPredict Function
- Generate the external model in Dataiku.
- Load the Dataiku model into a Vantage table.
- Use the DataikuPredict function to create predictions on the data stored in Vantage.
With Vantage BYOM DataRobot Function
- Generate the external model in DataRobot.
- Load the DataRobot model into a Vantage table.
- Use the DataRobotPredict function to create predictions on the data stored in Vantage.