To fully exploit AI’s potential across the enterprise, organizations should democratize DataRobot automated machine learning models in their existing analytics tools, processes, and apps. By bringing AI to your users, you can improve bottom line outcomes. In this series, I will show you a wide variety of options to infuse DataRobot machine learning models into your favorite analytics tools. Let’s dive in and see how we can put AI to work for everyone.
Delivering on the AI promise
One of the most common ways to responsibly democratize predictive insights is to add scoring into existing ETL / data pipelines and store results in databases for universal app and BI tool access. I personally prefer this approach whenever possible. In this scenario, you would add DataRobot into your organization’s existing governance, data security, change tracking, version control, and other trusted reporting processes.
In order to add DataRobot into your data pipeline, you’d first build a predictive model, select a deployment option, and then add the DataRobot-generated code into your ETL process. If you opt to use DataRobot’s REST API, you should be able to score data from just about any modern ETL tool. If you have Alteryx, modeling and scoring data is even easier. You can use the DataRobot Automodel and Predict tools.
In other cases, you might simply want to explore prediction model results and charts in your favorite data visualization tools. For those scenarios, you could use batch or live scoring with DataRobot API calls. The options are usually limited only by your own imagination.
DataRobot REST, R, Python, and other APIs
Any model built in DataRobot is production-ready with pre-built code to jump-start getting predictions. There is no need to write lengthy custom prediction code or manage infrastructure. Since most data preparation and analytics tools already support the use of R and Python, you might want to first explore adding DataRobot predictions or visualizations with DataRobot APIs.
DataRobot REST API in Microsoft Power BI Python Script Query Step
Regardless of whether you need real-time predictions — offline predictions, batch deployments, or scoring on Hadoop — you can easily operationalize machine learning models in production. Our Customer-Facing Data Scientists can help guide you through the many available API options. If you enjoy programming, you can add a reference to DataRobot’s library in Jupyter Notebook, R Studio, or your favorite scripting tool to dive in and see what is available.
DataRobot Python API
For those of you who don’t like to code, there are several simple no-code DataRobot third-party connectors and partner extensions you can use. Let’s explore those next.
DataRobot Excel Add-In
The first one I’ll walk through is the preview of DataRobot Excel add-in by CData. The CData Excel Add-In for DataRobot provides an easy way to batch score a CSV file with DataRobot right within Excel. After installing the add-in, you enter your DataRobot credentials, a file path, and then pick a deployed model to use for scoring. The predictions will be made and saved in your spreadsheet.
DataRobot Add-In for Excel by CData
In addition to scoring data, you can also use this add-in to get a list of projects, models, and deployed models. Note the CData add-in is currently only available for Windows environments. For Mac fans, please see the instructions for getting set up with ODBC on Mac.
DataRobot ODBC Connector by CData
ODBC/JDBC Connector for DataRobot
Another option for code-free batch scoring from a file with any ETL or business intelligence tool is the DataRobot OBDC/JDBC connector for any BI or ETL tool by CData. This option works with Windows, Mac, Linux, and Unix systems. In the image below, I’m showcasing the OBDC connector and inspirations from Nathan Patrick Taylor, Customer-Facing Data Scientist, that was publicly shared to the Microsoft Power BI Gallery.
DataRobot predictions used in Microsoft Power BI dashboard
Tableau + DataRobot
With Tableau, you can easily use DataRobot to efficiently focus on the right data to analyze, get predictive insights with explanations in your dashboards, and run simulations to get actionable prescriptive guidance on what to do next. The following example of a DataRobot AI-driven dashboard was shown in the DataRobot and Tableau in Action: A Zen Master’s View webinar. This dashboard is also available to interactively view on Tableau Public.
Tableau + DataRobot Readmission Prevention Dashboard by Teknion Data Solutions collaborators Will Grey, Joshua Milligan, and Bridget Cogley
Another use case for visualizing predictive models is to improve model performance. Here is another example from one of our Customer-Facing Data Scientists.
The DataRobot What-If extension takes analyses one step further by enabling Tableau users to experiment, simulate, and compare different scenarios using governed DataRobot predictive models to identify the best strategy or test ideas before committing resources.
DataRobot What-If Extension for Tableau
For more information and step-by-step tutorials for using Tableau with DataRobot, please sign up for the DataRobot Starter Kit for Tableau.
Qlik Extension: Qlik2DataRobot
Our newest partner extension led by Qlik is the Qlik2DataRobot connector for Qlik Sense. The Qlik2DataRobot server and client-side extensions enable Qlik users to ingest prepared Qlik data into a DataRobot project to create models. You can also use a script to score data in Qlik Load scripts to visualize predictions and create your own what-if scenarios using filters.
For More Information
That wraps up How To Democratize AI in Popular Analytics Tools Part 1. If you’d like to see more examples of AI-driven dashboards and live demos of popular democratization options, please watch the related series webinar.
In our next series article, I will share how to use the DataRobot API in Databricks , walk through Hadoop scoring, introduce MicroStrategy’s mstrio-py: data I/O for Python, and add DataRobot bots in UI Path and Automation Anywhere Robotic Process Automation (RPA) tools. If you have an idea for an integration or you would like to better understand existing DataRobot partner options, please let me know.
About the Author
Jen Underwood, Sr. Director Product Marketing, has held worldwide analytics product management roles at Microsoft and served as a technical lead for system implementation firms. She has experience launching new products and turning around failed projects in enterprise data warehousing, reporting, and advanced analytics. Today she designs products and helps analytics professionals learn how to solve complex problems with machine learning in the emerging citizen data science segment. Jen has a Bachelor of Business Administration – Marketing, Cum Laude from the University of Wisconsin, Milwaukee and a post-graduate certificate in Computer Science – Data Mining from the University of California, San Diego.