Making the jump from academia to industry
Advanced training in statistical thinking is an unfortunate exclusion from many scientific graduate programs. To resolve this skill gap—and better prepare a new generation of researchers for industry careers—BASF and KU Leuven jointly developed a course which identifies visual data analytics as a common ground between academia and industry.
There is a heavy emphasis placed on independent research in STEM-related graduate education. This structure relegates coursework beyond the scope of the dissertation material to secondary status. For instance, a Ph.D. candidate in organic chemistry is unlikely to be required to take advanced statistical courses, and vice versa.
While the objective of a graduate degree is to build domain expertise in a very specialized field, the lack of general statistical coursework can be a roadblock to newly-minted graduates seeking employment in industry.
Simply put, there is a disconnect between what students are learning and what they need to know to succeed in non-academic sectors.
No blame need be placed in this conundrum; after all, industry and academia operate under different paradigms. While industry is fast-paced and focused on maximizing results, academia adheres to precedent and focuses on fundamental understanding. Both models serve necessary niches. However, with the U.S. National Science Foundation estimating 47% of scientific graduate students opt for careers in industry, it is imperative that academic institutions better facilitate students’ transition from one sector to the other.
Point-and-click analytics as a common ground
Enter JMP. Focused on democratizing statistics and making analytics intuitive, JMP Statistical Discovery software is uniquely suited to bridge the chasm between academia and industry. While script-based tools such as R and Python are capable and popular, their steep learning curve forms a roadblock to deriving meaningful information from raw data. Conversely, JMP allows for point-and-click data investigation for maximal efficiency in analytics and experimental design. This ease makes for smoother integration into graduate education, where, for example, chemical engineers may not have time to devote to learning programming.
It follows that statistical thinking forms the common ground between academia and industry—a truism that leading chemical manufacturer BASF has not only recognized but acted upon.
At KU Leuven, the company now offers an introductory statistical analytics course for master’s students in chemical engineering. What began as a short summer class quickly evolved into a full academic-year offering, confirming that students have keen interest in data literacy.
"We are fitting statistics into the DNA of chemical engineers so that when they graduate, they can troubleshoot on their own before involving [their employer’s] data analytics team."
Data Analytics Specialist, BASF
Designing a course around statistical skills in any software confers myriad benefits to students, and, Stuyck says, there were reasons for choosing JMP in the design of this class. Namely, all ten of the world’s largest chemical companies—including BASF—use JMP. KU Leuven graduates already experienced with JMP would therefore possess a readily applicable skill should they opt for careers in industry.
Data literacy far transcends the classroom
As data science and programming have pervaded scientific industry, the idea that data literacy provides benefits to the employer has, in prior years, been broadly—but uncritically—accepted. Deeper analysis of that principle, however, begs the question of what exactly data literacy entails. What does it mean to be data literate? Stuyck outlines three main qualities which highlight the importance of statistical education.
1. Independent Troubleshooting. Frequently in industry, there are designated statistics teams to which engineers will hand off data for analysis. By strengthening analytic thinking across the organization, BASF can quell the inefficiencies of outsourcing analysis while maintaining higher data fidelity along the way. "
There is a new focus on increasing data literacy among chemical engineers and process managers,” Stuyck says. JMP, he explains, is a code-free, point-and-click solution that makes analysis approachable.
"JMP comes in very handy since you can do a lot with it, but you don't need a lot of programming background, which is not always within the scope of education in chemical engineering."
2. Data Exploration. Scientists and engineers must be able to dive headfirst into data, even when they don’t know what they’re looking for. Visual and interactive data exploration confers not only data literacy, but analytic confidence, Stuyck argues. Further, he adds, JMP plays another critical role here: “If engineers are struggling with a tool or programming language just to visualize data, then the solution set will be suboptimal.”
In contrast to script-based solutions, JMP’s emphasis on visualizing data makes exploration and understanding accessible, code-free, even to a statistical novice.
That learning curve is further smoothened by a large collection of free educational resources available on JMP.com. “JMP has some very nice tutorials and videos on how to perform tasks,” Stuyck says, explaining that this repository allows users to quickly get out of a software rut and focus on scouring the data for trends.
“It is just so refreshing to see how things are done, and to identify traps which need to be circumvented,” says former student Richard ten Hagen.
"I appreciated the practical perspectives on problems from non-professor experts."
Richard ten Hagen
3. Data Confidence. Stuyck explains that constructing an attitude of data-driven decision-making is crucial to building an analytic workflow. This idea was a central tenet of BASF’s course at KU Leuven, where case studies made for a “learning by doing” environment, says Ten Hagen. The course’s hands-on approach embraced data from beyond the chemical field, strengthening students’ data-driven critical thinking skills.
“The BASF course helped in appreciating the work and difficulty of applying models to complex industry issues,” says student Toon Verhaegen, adding that students also enter the workforce more aware of the domain challenges.
Together, these skills mesh to create a cohort of graduates both fluent in and confident with data, capable of piloting the entire data life cycle from experimental design to troubleshooting to final decisioning. JMP sits alongside these scientists, facilitating each step of their data literacy journey – not just at university but in their careers beyond.
Streamlining research in the face of growing competition
While many STEM graduate students later enter the industrial workforce, the National Science Foundation estimates 44% will stay in academia. Fortunately, academic institutions also stand to gain from a workforce skilled in the advanced statistics needed to speed scientific discovery while also reducing the cost of experimentation.
For example, design of experiments (DOE)—a statistical approach widely accepted as the gold standard of lean experimentation—could doubly benefit research labs by paring down both the time and resources needed to execute an experiment. Industry benchmarks indicate that the transition to DOE has been accompanied by 50-70% savings in time and cost. In a sector where p = 0.05 is a perennial standard, assessing statistical significance with lower expense could prove transformational.
Academic research churns out fundamental discoveries and novel characterizations, and yet remains woefully underfunded. Principal investigators often rely on external grants to finance their work, and grant rejection rates continue to climb as federal research funding stagnates. Moreover, interdisciplinary research grant proposals are even less likely to be greenlit, further diminishing the ability for top minds to collaborate.
A reduction in overhead laboratory costs could allay, at least temporarily, some of the headaches generated by this funding scheme. With the ability to stretch grant dollars farther thanks to DOE’s cost-effectiveness, principal investigators can enjoy fewer administrative headaches. Additionally, the National Science Foundation, an enormous disburser of grants, is increasingly placing value on sustainability. A grant proposal with a very lean approach to experimentation would generate less waste, likely giving it an edge in securing funds.
This potential explains why stakeholders at KU Leuven—and BASF—have invested in teaching cutting-edge statistical methods like DOE. In essence, a new generation of students skilled in the applications of designed experiments and capable of executing those methods in JMP are positioned to advance scientific discovery regardless of whether that discovery occurs in academia or industry.
A strategic win-win
In short, there is a demonstrable need for robust connection between academia and industry, and JMP a natural bridge. Democratizing statistical thinking is most straightforward when roadblocks and headaches are small. JMP, with its intuitive graphical interface and robust support library, smoothens the learning curve and encourages users to ask “what,” “why,” and “how” about their data.
“I have already used JMP on another project, and it is now part of my toolbox,” says ten Hagen. BASF took the first step towards closing the gap between academic and industrial skillsets, and their efforts could represent a shift towards more comprehensive science and engineering graduate education.
Everyone wins from BASF’s investment: KU Leuven can offer students access to innovative pedagogy designed to help them succeed after graduation. Students are equipped with the statistical skills needed to solve the pressing scientific challenges of our time—and do so with fewer resources. Lastly, corporations like BASF are tapping a talent pool of independent, curious analysts who are ready to contribute on Day 1. It’s another example of how democratizing statistics pays dividends to all.