Capstone Projects

John Deere Expert Alert Finder: Unsupervised Approach to Discovering Expert Alerts

Program: Data Science Master's
Host Company: John Deere
Location: Moline, Illinois (onsite)
Student: Robert Margherio and Chris Spaude

Expert Alerts are a proactive way to provide John Deere customers with the best in class service. They are one of the multiple tools enabled by John Deere Connected SupportTM. Creating the algorithms, or rules, for Expert Alerts is a time-consuming process. Generally, employees manually mine through data from machines to determine how to identify an issue. This analysis utilizes unsupervised machine learning algorithms to mine for potential Expert Alerts.

Multiple data sources, internal to John Deere, are wrangled and combined to create a dataset ready for analytics. Two unsupervised machine learning methods, market basket, and network analysis were applied to find valuable Expert Alerts. The algorithms work by finding patterns between DTCs and part numbers. Our end-users, the Expert Alert developers, have the ability to mine results through two separate but related interactive reports.

The outputs from the algorithms need to be checked before moving into production. We allowed testing time and have initial end-user feedback of the two reports from a sample of our Expert Alert developers. It is up to our end users to see value in the data set and reports we created, which should help produce more Expert Alerts for our customers.

“UW’s Masters of Science in Data Science program gave me the experience and knowledge I needed to expand my skills in a career in data and analytics. The capstone project specifically gave me an avenue to explore a real-world problem, of my own choosing, to demonstrate my newly acquired skills.” – Robert

“This data science program was able to provide me with the tools and techniques that were directly applicable in my data science job. Being able to take what I have learned in a class and apply it to real-world applications was very beneficial in my learning.” – Chris