My elderly mother called me up last week to tell me how she had learned all about Artificial Intelligence (AI) from a feature on the BBC Radio 4 Today show. She's an expert now! I gave her a healthcare demo of DataRobot, an automated machine learning platform, showing how easy it is to make predictions from data — predicting hospital patient readmissions somehow seemed relevant to her situation.
Similarly, my twenty-year-old daughter who is studying to become a veterinarian was testing horses by putting them on treadmills, turning up the speed levels, and recording the results. After watching her struggle with data analysis in Excel, I loaded the data into DataRobot to see what predictions it would make. Sure enough, the results were instant and impressive — predicting the unfortunate outcome of turning the treadmills up to number eleven!
I started in IT at Oracle in 1988, and for the first time in my entire career, my mother and daughter both understand the impact of the technology I am marketing. Surely when millennials and octogenarians are aligned on technology, it’s is a strong sign that you’ve reached a mainstream market!
However, as with any transformative technology, buzz and hype abound as everyone jumps on the bandwagon. To bring clarity, we turned to best-selling author and President’s Distinguished Professor in Management and IT at Babson College, Tom Davenport, to quantify the benefits. His white paper sets the context of AutoML and includes specific use cases where organizations are putting it into action.
For his research, Davenport interviewed real companies using AutoML and below are just two of the many benefits being realized by organizations that are becoming AI-driven enterprises.
1. Modeling Productivity & Effectiveness
Creating models is a core process for machine learning and involves various activities such as feature (variable) selection and engineering, data preparation, selection of algorithms, and evaluation and comparison of results. Each step alone takes time when tackled manually, which is an inefficient use of time.
Automated machine learning technology replaces these manual processes resulting in increased productivity and effectiveness of the models. Automation ensures that human errors aren’t easily made, and enables data scientists and non-data scientists alike to deploy models and test them out, rather than spending time trudging through each manual process.
Sumitomo Mitsui Card Company (the largest credit card company in Japan) uses DataRobot for risk modeling and customer insight/marketing applications. Davenport references this company in his research about how automation enables them to build and validate models in hours to days, rather than months:
“In the risk modeling area, some analysts were doing machine learning manually, but it could take up to half a year to build and validate a model. Using DataRobot cut that time to hours or a few days.” - Sumitomo Mitsui Card Company
2. Reduced Skills Requirement
Data is everywhere, but not everyone knows how to use it to resolve business problems. (Will this applicant default on a loan? Should I order more of this product for the holiday season?) Automation lets non-data scientists (such as expert business analytics professionals) find the best models to solve their specific business challenges. The democratization of data science is achieved through automation so that both data scientists AND non-data scientists can leverage their data and see results.
The demand for data scientists is also greatly surpassing the supply which means there aren’t enough data scientists to go around, and not every business has the budget to take on a data science expert. As more businesses adopt machine learning, this gap will only continue to grow. Davenport acknowledges this shortage in his research:
“It is widely known that there is a shortage of advanced skills, and since AutoML does a lot of the expertise-based tasks in machine learning [...] it can help alleviate the skill shortage.” - Thomas Davenport
Interested in seeing the complete list of benefits from automated machine learning? Read Tom Davenport’s white paper here. Or talk to my mother.
About the Author:
Tim Young is responsible for global marketing at DataRobot. He has over 25 years of experience marketing high-tech enterprise products in the data management, analytics, and SaaS spaces. He has run marketing operations for global companies including IBM, Oracle, Netezza, and Workday. Tim brings a practical global perspective to DataRobot having managed marketing teams in Australia, Asia, U.K., Europe, and the U.S.