Using JDBC to connect to Incorta SQL Interface from DataRobot

[suxinji]

• Overview
• What you should know before reading this article
• Solution
• Create New Project
• Select Data Source
• Create JDBC Connection
• Best Practices
• Only Share the Schemas Required
• Create Business Schemas
• Reuse and Share the Data across Multiple ML Projects
• Reference

 

Overview

In this article, we will show the steps that are required to leverage data prepared in Incorta as input for modeling, building and inference in DataRobot.

What you should know before reading this article

• About Incorta SQL Interface
• SQLi: Understanding and Best Practices

Solution

Here are the step by step instructions of creating a connection from DataRobot to Incorta.

Create New Project

Select Data Source

Create JDBC Connection

Select the postgreSQL database as the data store.  

Best Practices

Below are some best practices when using Incorta as the source for data input for AI/ML platforms, such as DataRobot.

Only Share the Schemas Required

Incorta allows you to control who can have access to a schema.  A schema or a business schema can be shared with View only access which is generally sufficient for the consuming team to access the data.

Create Business Schemas

Both Incorta business schemas and physical schemas can be exposed and selected from the AI/ML platforms such as DataRobot.  However, it is easier for the consumer to uptake the data from Incorta when the data is already joined and a flattened view is provided.

Reuse and Share the Data across Multiple ML Projects

You should treat Incorta as the "Feature Store" to maximize the reusability.  Incorta business schemas and business views provides high level data abstraction.  This enables data scientists to focus on the ML model selection and optimization, and leave the data preparation to the data engineering team using Incorta.   Data can be joined, aggregated, and consolidated all within Incorta.

Reference

DataRobot documentation - DataRobot data connections