<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=95044&amp;fmt=gif">
   
Back to blog

A Conversation About Machine Learning

Wednesday, September 05, 2018

Chris Wintermeyer of Big Squid spoke with Kraken customer evangelist Jennah Crotts of the University of Southern California (USC) about the challenges of predicting student enrollment and attrition and the vital role of machine learning. 

 

In their conversation, Chris and Jenah address questions such as:

  • How do you decide when you need predictive analytics?
  • How is life different from predictive versus descriptive analytics?
  • What’s the onboarding process like for an automated machine learning platform like Big Squid’s Kraken?
  • How do you measure ROI associated with a machine learning platform?

Crotts, Assistant Director of Admissions Operations and Strategy at the largest US school of social work and center of a massive alumni network, is keenly interested in enabling people within her organization to do smart things with data faster and more efficiently.


While higher education is a bit different from other sectors, UCS’s leadership wants to see data drive decision-making as much as the leadership in any other organization. A sales funnel becomes an admissions funnel reliant on good lead acquisition: if they can accurately predict which students are unlikely to complete an application, they can do active outreach to support that student in her application process.


USC came to Kraken through its use of Domo. They were happy to get a good visualization tool, but found that it was only really as valuable as their ability to accurately predict what would happen six months down the line. They faced a roadblock many organizations encounter as they analyze historical data: are outcomes different this year because of something they’d done or because this year was just different? They simply didn’t have the capacity to understand what was behind the descriptive analytics they were seeing.

And for that they needed Kraken.

Implementing Kraken was seamless—even for Crotts, who usually dislikes bringing in outside organizations and tools. As she points out, she works with many smart people in higher education and wants to enable them to do better rather than bringing in extra tools and third parties.

But she found collaborating with Big Squid fundamentally different from experiences she’s had with other organizations. Using Kraken is about enabling the people you already have on site. They don’t necessarily need to have a real strong stats background or be amazing at manipulating data or writing code.

Big Squid and the Kraken platform build on the skills your team already has and help you understand your business better. It empowers those with business acumen but not extensive stats or coding experience to run predictive scenarios and increase the value they extract from their data. Kraken is accessible to everyone.

The implementation process is really just a learning process to help you identify goals and migrate datasets into the predictive analytics platform.

Throughout the conversation, both Crotts and Wintermeyer emphasized that working with Kraken is an iterative process. “You’re not going to get a perfect model right out of the gate. You refine models as you see the predictions that are coming out of it,” Wintermeyer noted.

After a year of working with Kraken, Crotts is excited about spreading the news about the impact Kraken’s predictive analytics can have throughout USC’s other 16 schools. “We’re telling our various departments to be data-driven decision makers, but we haven’t really been giving them the tools to do that,” she said.

It’s become a mission of hers to get Kraken in front of them so that they can see for themselves how the platform can drive results.

Learn more about how Kraken’s automated machine learning platform can help your company accelerate your data value journey and improve outcomes.

Big Squid

Written by Big Squid

Recent Posts

Comment Below

Subscribe to Email Updates