John Szarka
Statistician, W. L. Gore
Meg Hermes
User Reference Manager, JMP
Statistician John Szarka supports regional and global new product development and process improvement for W. L. Gore & Associates’ Core Technology group. His work integrates many areas of industrial statistics, including design and analysis of experiments, statistical process control, measurement systems analysis and sampling plan developments. He is a champion and core team member for Gore’s analytics initiatives establishing best practices and training curriculums across the enterprise.
JMP User Reference Manager Meg Hermes spoke with John about how W. L. Gore has cultivated a dynamic analytics culture, standardizing around JMP Pro and engaging with all the available resources to keep learning and improving.
Meg: Many people think of W. L. Gore as something of a pioneer in analytics transformation, given the company’s long history with data-driven approaches. But keeping that analytics culture in a state of evolution is no doubt an ongoing process. What kinds of ROI or improvements have helped you and your colleagues to sustain momentum – and keep innovating – when it comes to analytics at Gore?
John: I feel very fortunate that statistics and analytics are rooted in the founding of our enterprise. Our co-founder, Bill Gore, wrote the book Statistical Methods for Chemical Experimentation back in 1952 while still at DuPont, before the company was started in 1958. Our investment in having a global statistics team has been highly valued by our technical leaders across the organization.
With such an explosion of "big data" and modeling advances, one thing I feel we’ve recently taken very seriously at Gore is valuing our data as an asset. Having the right people and systems in place to enable us to extract the most value from our data is essential for us now and moving forward.
We want to ensure that we are using the right data to answer the right questions. That remains at the front end of our conversations when we’re exploring projects using big data for new product development or process improvement studies. In many instances, we can use JMP Pro to meet our needs on the data and analysis side to provide insights and help in making decisions. Our investment in JMP Pro – up from the base version of JMP – has made this type of modeling much easier for our associates.
Meg: One of the most common objections we hear from newcomers to JMP is that some fear statistical approaches will supplant domain expertise. How has having JMP Pro changed the way domain experts at Gore apply their knowledge and skills?
John: It’s never an either-or argument when it comes to using data from the process versus domain expertise. We make the best combined inferences when we bring this information together. There is a strong culture of experimentation at Gore. We may have information from the past around how we think a certain experiment may work out. In some instances, the actual data agree with those prior beliefs, and in other instances they don’t.
One of the key points is to remind the group around the reproducibility of the experiment. We are likely to build on the work of one experiment with a follow-up experiment. And if we continue to see a particular signal, we can be more certain that it’s true. If it falls off, then perhaps not. Over time we build robust knowledge from the science and the data together.
Meg: Tell us about your relationship with the JMP organization. What value do you get from participating in Discovery Summits, users’ groups, training, STIPS, the online JMP Community, etc.?
John: The online JMP Community is a vibrant space that I try to visit at least once a week. It has been a great place to learn new tips, read discussions on various issues and get a sense for how others are leveraging JMP for their organizations. I’ve probably contributed the most myself in the "wish list" category – mostly based on observations that could improve the JMP experience for our users at Gore even more.
This past year, we as a statistics group at Gore looked at modifying our in-person trainings for the virtual environment. Utilizing STIPS has been great for us because it gives rich context in the type of problems our technical associates are trying to solve. The content is framed with the user in mind – how they can answer their questions using statistical techniques. Many software vendors would have a series of "click here" steps that you follow, but I was really impressed with the production quality of many of the STIPS modules.
The online Discovery Summit conferences in the Americas 2020 and Europe 2021 were the first I attended. I gravitated to the presentations from industry, as the problems they were working on were interesting. One of my favorite presentations was Statistical Process Control for Process Variables that have a Functional Form. Profile monitoring in the SPC literature has existed for some time, but I really liked how easily we can use platforms like Functional Data Explorer and the Model Driven Multivariate Control Chart to easily implement them in practice.
Meg: Tell us about your relationship to the JMP software developers. How do you go about providing feedback, and do you feel that feedback helps influence new JMP releases? Does your ability to maintain a collaborative relationship with JMP Development have an impact ultimately on the value you get from JMP?
John: The JMP software developers are a fantastic group of people to interact with! I’ve made connections with JMP associates over the years through external conferences along with JMP Discovery Summit. That led me to use the Early Adopter program to get a sense of the direction of future releases and have a voice into that process.
The developers are great listeners and really seek to understand our problems and the solutions that will help us the most in future JMP releases. This has also led to joint meetings between members of Gore and JMP to discuss targeted areas. I’ve seen multiple instances where specific requests I’ve made have been implemented, which is proof that this process works.
Meg: What advice would you give to someone who is just starting out with JMP?
John: Partner with those in your organization already using JMP or reach out to your peers in quality organizations you’re a part of to see the areas they get the most value out of JMP. I would personally start in the graphing menus, especially Graph Builder. There is tremendous power in storytelling with graphs and I find that I can tell a lot of stories by utilizing the many features within Graph Builder. As many users start learning more about the various models you can build in JMP, they sometimes get away from being as good at graphing, but I would get into the habit of centering your focus on graphing from the onset.