Teradata Vantage Analytics Capabilities
Teradata Vantage™ is a data and analytics platform that offers a wide range of features for data preparation, feature engineering, hypothesis testing, analytic modeling and scoring. It accelerates analytic workflows by providing a highly performant, scalable, enterprise data platform that performs production-advanced analytics with little to no data movement to and from other platforms. Teradata Vantage offers the best-of-breed choices and the ability to conduct analytics using tools and languages preferred by analytics professions such as SQL, Python, R and third-party tools Dataiku, H2O, including any PMML-compliant tool.
- Vantage Native Data Preparation, Transformation and Feature Engineering
- Analytic Database Analytic Functions
- Unbounded Array Framework Tme Series
- Vantage Analytics Library (VAL) for descriptive statistics, data transformation and reorganization, hypothesis testing, model building and model scoring functions
- Vantage Open Analytics: R and Python
- Script Table Operator and Exec R Table Operator
- Script Table Operator Orange Book (log on required)
Orange Book: R and Python Analytics with SCRIPT Table Operator
- Vantage Open Analytics: Bring Your Own Model (BYOM)
- Predictive Model Markup Language (PMML) model scoring
- H2O Model Object Optimized (MOJO) model scoring
- Vantage Open Analytics Client Libraries
- Teradata Package for Python
- Teradata Package for R
Vantage implements its native and VAL analytic capabilities as SQL functions. Then, adds Python and R interfaces on top of that within the client libraries. These client libraries are generating and performing the appropriate SQL syntax for the analytic function transparently to the end-user. Data scientists are using R and Python and benefiting from the performance and scalability provided by Teradata Vantage.
Vantage allows data scientists to run R and Python natively in the SQL Engine using table operators. The client libraries provide R and Python interfaces to generate the SQL syntax required transparently to the end user. If you have already selected a different analytic platform, the client libraries provide standard R and Python methods of exporting and loading data in parallel, not serially. Once your modeling is done in the other platform, PMML and H2O MOJO can be used to score the entire database back in the Vantage platform using SQL or the interfaces from the R and Python libraries.
Everything Vantage does from an analytic perspective can be accomplished in SQL, R or Python.
Data Exploration, Data Cleaning, and Feature Engineering Functions
Modeling Algorithms, Model Training, and Model Evaluation Functions
| Type of Function | SQL Reference | Python Reference | R Reference |
|---|---|---|---|
| Vantage Analytic Library Algorithms | SQL: Vantage Analytic Library Algorithms | Python: Vantage Analytic Library Algorithms | R: Vantage Analytic Library Algorithms |
| R & Python Micromodeling | SQL: R & Python Micromodeling Log on required |
Python: R and Python Micromodeling | R: R and Python Micromodeling |
Model Scoring, Model Monitoring, and Life Cycle Management Functions
| Type of Function | SQL Reference | Python Reference | R Reference |
|---|---|---|---|
| Vantage Analytics Library Library Scoring | SQL: Vantage Analytic Library Scoring | Python: Vantage Analytic Library Scoring | R: Vantage Analytic Library Scoring |
| R and Python Model Scoring | SQL: R & Python Model Scoring Log on required |
Python: R & Python Model Scoring | R: R & Python Model Scoring |
| BYOM Scoring | SQL: BYOM Scoring | Python: BYOM Scoring | R: BYOM Scoring |
Machine Learning Engine
The Machine Learning and Graph Engines (ML/G) provide extensive analytic capabilities. The Teradata Vantage™ - Machine Learning Engine Analytic Function Reference provides syntax and examples for all its functions. The table below gives links to the various sections.
| Type of Function | SQL | Teradata Package for Python | Teradata Package for R |
|---|---|---|---|
| Statistical Analysis | Statistical Analysis | Teradata Package for Python Function Reference | Teradata Package for R Function Reference |
| Path and Pattern Analysis | Path and Pattern Analysis | ||
| Data Preparation and Transformation | Data Preparation and Transformation | ||
| Cluster Analysis | Cluster Analysis | ||
| Time Series Analysis | Time Series Analysis | ||
| Predictive Modeling | Predictive Modeling | ||
| Text Analysis | Text Analysis | ||
| Graph Analysis | Graph Analysis | ||
| Association and Recommendations | Association and Recommendations |