In anticipation of his upcoming presentation at Predictive Analytics World for Industry 4.0, Las Vegas, June 19-24, 2022, we asked James Duarte, Principle, IMAGILYTICS and Academician at IMAGILYTICS, a few questions about their deployment of predictive analytics. Catch a glimpse of his presentation, Machine Learning and Data Science Value Chain, and see what’s in store at the PAW Industry 4.0 conference.
Q: In your work with predictive analytics, what behavior or outcome do your models predict?
A: My predictive analytics work has covered process performance and outcome improvement. Predictive analytics work covers the use of both structured and unstructured data, including disambiguation techniques.
Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?
A: Predictive analytics are part of the “Data Science Value Chain” that not only delivers results, but also maximizes the understanding and effectiveness ensuring “good data” for modelers that yields better decision-making. By using a cogent example, the differences between designed experiments and machine learning for finding best results becomes clear.
Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?
A: Predictive analytics contributed to many activities. One example is reducing raw material costs by $750k per year. By combining data mining with purchasing relationships and trends the organization maximized purchasing decisions that saved $350M over a 7 year period.
Q: What surprising discovery or insight have you unearthed in your data?
A: A quality issue occurred in rolling aluminum can stock. Every problem solving tool was used to no avail. Machine Learning discovered the “hidden variable” that engineers and metallurgists refused to accept until one of their own explained how it was possible then controlling it allowed the company to “take the market” for 10+ years.
Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.
A: The prime take-aways are three-fold: 1) revealing the keys to obtaining “good data” for machine learning and modelers, 2) clearly defining data science roles to enable organizations to maximize their investment in data science personnel, and 3) combining the right machine learning tools to solve the problem. Machine learning can’t be successful without good data, the right people in place and knowledge of the right tool(s) with software to support it.
Q: How can predictive analytics be successful without good data and the right people performing the right tasks?
A: Good data has four elements: 1) proper collection, 2) proper storage (if “data at rest”, as opposed to “data in motion” from sensors), 3) properly cleansed, and 4) accessible to those who need it. There is no “one person” that can perform the roles of computer scientist, advanced analytics modeler, business analyst and senior management liaison to ensure the best decision-making process. The roles are complementary and overlap thus they must be part of a collaborative across the organization, not held in one place.
By: Steven Ramirez, Conference Chair, Predictive Analytics World for Industry 4.0