The Gartner Data and Analytics Summit was held March 18-21, 2019 in Orlando, FL. The excitement generated from the over 4100 participants was evident even at the start of the opening keynote, “Lead With Purpose to Achieve Clarity in a World of Ambiguity”.
The 2019 U.S Data and Analytic Keynote Team including Donald Feinberg, Me, Valerie Logan and Michael Moran
To be honest, the excitement began BEFORE the opening keynote with both the BI Bake-Off and the Innovative Analytics in Action sessions, two very popular, exciting sessions that kicked off on Sunday and were led by Cindi Howson and Rita Sallam, respectively.
But this year, we pushed the envelope even further – adding the new Data Science and Machine Learning (DSML) Bake-Off as well as Show Floor Showdowns to provide additional opportunity for participants to see analytics tools in action across the analytic capability spectrum AND across the end-to-end analytic process pipeline.
The DSML Bake-Off showcased three participants this year: Sean Owen presented Databricks, Colin Priest demonstrated DataRobot and Jonathan Wexler demoed SAS. The presenters did a fabulous job showcasing and contrasting their distinct approaches for solving the same business problem.
The DSML Bake-Off in Action
The session was attended by over 350 participants! Whew – I’m thankful we proactively moved the session to a big room that could accommodate more than 50 people, which is what we initially anticipated!
Our objective for the session was to demonstrate data science and machine learning vendors that define the various clusters represented in the DSML market today – “apples and oranges”, so to speak – and outlined in the research “How to Choose the Right Data Science and Machine Platform”. In order to ensure that the demos remained educational and did not become pure marketing vehicles, vendor participation was based on a combination of customer interest, market momentum and product maturity. We considered vendors based on three different product approaches for three different use cases (aka “clusters”):
- Notebook-based/Programming approach – focus on developers and experts wanting to leverage a broad range of open source and other tools to create and operationalize models.
- Pipeline/Workflow analytics approach – focus on expert data scientists with a stepped and canvas interface approach for working through the analytics process from data access to operationalization.
- Augmented approach – focus on non-expert, citizen data scientists needing an automated and augmented approach to leverage predictive capability as well as facilitation to work collaboratively with data scientists.
The live demo topics focused on key user requirements in which vendors often take a different approach. The script and data sets were given to the participants in advance. This year, we choose the College Scorecard data and focused on determining the best predictors of schools most likely to have successful students 10 years after graduation. The session consisted of each vendor demonstrating the use of their platform for each of the tasks within the script. We moved through the tasks, one by one, each vendor showing their capabilities for each task before moving to the next. Polls conducted after each section enabled participants to engage and provided vendors with immediate feedback on articulating product capabilities. The tasks demonstrated included accessing and prepping the data (including feature selection), model building, training and validation, model delivery and management, and delivery of final business results and recommendations.
The differences among the platforms were immediately obvious and stark. Here’s a taste of what was demonstrated from each of the vendors:
SAS demonstrated their modernized approach aimed at leading data scientists through the end-to-end analytic workflow.
SAS VDDML Feature Engineering and Model Pipeline including Open Source
DataRobot focused more on the citizen data scientist who wants to leverage advanced analytic capability but is not a coder nor familiar with the DSML process or the expert who wants to automate more repetitive DSML tasks.
DataRobot Drills Down into Model Validation
Databricks demonstrated the ability for developers and experts to collaborate, to leverage their knowledge and experience and to orchestrate the analytic process across a wide breadth of data and tools, including open source, at scale.
Databricks MLflow for Tracking and Managing Models
Be sure to check out the newly published Gartner research “Critical Capabilities for Data Science and Machine Learning Platforms” to dive into the additional capabilities provided by each of these vendors.
The Show Floor Showdowns provided a nice complement to the Bake-Offs. The Showdowns were conducted on the vendor show floor and consisted of six analytic and BI (ABI) vendors and five DSML vendors.
What a wonderful crew we had to demo the platforms for the Showdowns! For the ABI sessions, the participants were Ziad Fayad for Salesforce Einstein Analytics, Pete Reilly for AnswerRocket, Mary Flynn for ZoomData, Vandita Manyam for Tellius, Vicky Lozovsky for Information Builders WebFOCUS and Brad Hopper for Tibco Spotfire. For the DSML sessions, the participants were Ainesh Pandey for IBM Watson Studio, Feng Bai for H2O Driverless AI, Jorge Zuloaga for Big Squid Kraken, Ryohei Fujimaki for NEC DotData and Ingo Mierswa for RapidMiner.
Each presenter presented their platforms by including a live voiceover over a 10 minute pre-recorded demo using scripts and data consistent with the Bake-Offs (College Scorecard data for DSML and Loneliness data from the Kaiser Family Foundation and the Economist for ABI). To be selected for these sessions, vendors from the show floor and the corresponding MQs first submitted a form indicating their interest. From the pool of over 60 applicants, the final participants were randomly selected. Unlike the Bake-Offs, the demos for the Showdowns ran through the complete analytic process, start to finish, and there was no voting by attendees.
One insight I found particularly interesting during the ABI Showdowns was AnswerRocket’s finding that people from Ohio tend to be the loneliest! Being a native from the Buckeye state, this made me feel so alone! 🙁 Haha.. Not really… 🙂 But I did wonder out loud if the survey was conducted in February, a time of year when most Ohioans may be feeling a bit isolated. At the other end of the scale was Florida – further making me wonder…
But even the beauty of the snowy, cold Ohio winter can make me smile! Photo compliments of Jonathan Geiger.
About 200 participants attended both the ABI and DSML sessions and, like the DSML bake-off, feedback again was extremely positive. Participants appreciated being able to see the platforms side-by-side, walking through a typical analytic workflow.
Thanks to all the vendors and the participants for helping us to take this next big step in providing sessions that share the state of the analytics market by enabling us to dive in – up close – in a fun and exciting way.
And, what’s more, we’re already working through the feedback and looking forward to honing these sessions for future Gartner events!
Disclaimer: Data from these analyses are for demonstration purposes only.
View Free, Relevant Gartner Research
Gartner's research helps you cut through the complexity and deliver the knowledge you need to make the right decisions quickly, and with confidence.Read Free Gartner Research
Comments or opinions expressed on this blog are those of the individual contributors only, and do not necessarily represent the views of Gartner, Inc. or its management. Readers may copy and redistribute blog postings on other blogs, or otherwise for private, non-commercial or journalistic purposes, with attribution to Gartner. This content may not be used for any other purposes in any other formats or media. The content on this blog is provided on an "as-is" basis. Gartner shall not be liable for any damages whatsoever arising out of the content or use of this blog.