To better understand functional data analysis (FDA) and its practical applications, I found myself delving into online research thanks to ScienceDirect and Google Scholar, in search of relevant examples. I was fascinated to discover the incredible diversity in how scientists and researchers are using the FDA features in JMP Pro to further their scientific discoveries.

When pharmaceutical scientists in Portugal, NASA engineers in Virginia and Alabama, chemical process experts at Dow, sustainability researchers in Arkansas, and adhesive formulators at Henkel Corporation in Connecticut all independently choose the same analytical approach, something significant is happening.

These teams, working on vastly different challenges, discovered that traditional data analysis was leaving critical insights on the table, and FDA was the solution.

When standard methods fall short

Across industries, scientists and engineers face a common frustration: their instruments generate rich, continuous data such as dissolution curves showing how drugs release over time, thermal profiles revealing material transitions, spectroscopic signatures capturing molecular behavior, but conventional analysis methods reduce this wealth of information to a handful of discrete points or peak heights.

This disconnect between measurement and analysis creates blind spots. Important patterns hidden between sampling points go undetected. Subtle but meaningful variations in curve shape get lost. And when you're developing life-saving pharmaceuticals, ensuring rocket motor safety, or maintaining production quality, these blind spots matter.

Functional data analysis offers a fundamentally different approach: treat the entire curve as the data point. Instead of reducing a dissolution profile to measurements at 60, 120, and 180 minutes, analyze the complete functional behavior. Instead of comparing peak heights in spectra, examine the full spectral fingerprint.

The results? Five research teams, five different industries, one consistent conclusion: FDA reveals what traditional methods miss.

Accelerating drug formulation in pharmaceutical development

The challenge

Researchers at Universidade de Coimbra and Grupo Tecnimede in Portugal faced a time-consuming bottleneck in developing extended-release tablets. Traditional dissolution testing requires processing each time point independently, which makes it difficult to predict how formulation changes would affect the complete release profile. As they noted in their study, classical design of experiments "does not account for the dynamic nature of drug release kinetics."

The FDA solution

The team applied functional design of experiments (DOE) using JMP Pro to model dissolution profiles as full curves rather than isolated time points. This allowed them to evaluate how formulation changes influenced the entire drug release process at once. The result: predictions as accurate as neural networks, with the added benefit of clear, interpretable insights into why certain formulations performed better.

Ensuring rocket motor safety in aerospace engineering

The challenge

NASA's Langley Research Center in Hampton, VA, and the Marshall Space Flight Center in Huntsville, AL, needed to characterize thermal insulation performance in solid rocket motors, which is a critical safety concern. With 43 test cases and 128 measurement points per test, the team had rich data but complex relationships. Material decomposition rates varied continuously along each test sample, influenced by multiple environmental factors including oxygen flow rate, pressure, temperature, and alumina content.

The FDA solution

NASA engineers used JMP Pro to treat thermal decomposition profiles as functional data, applying functional principal component analysis (FPCA) and functional regression to see what truly drove insulation performance. This made it clear which environmental factors mattered most and enabled reliable predictions under new conditions. With those insights, the team could evaluate future rocket motor designs without the cost and delay of exhaustive physical testing.

Preventing equipment failures via chemical process safety

The challenge

Researchers from the University of Coimbra and The Dow Chemical Company tackled a pervasive problem in chemical manufacturing: fouling in heat exchangers during multiproduct batch processes. The complexity was substantial: different batches used different recipes with varying operating conditions, raw materials, and process parameters, making traditional batch-to-batch analysis extremely difficult.

The FDA solution

Using FPCA in JMP Pro, the team transformed complex batch process data into functional features that could be compared across recipes and operating conditions. These features exposed clear, monotonic trends tied to fouling behavior—well before failures occurred. By pinpointing pressure drop as the strongest early indicator, the researchers established practical thresholds that enabled proactive maintenance and safer, more reliable operation.

From waste to wax: Upcycling bioethanol byproducts

The challenge

Researchers at the University of Arkansas saw untapped potential in a common bioethanol byproduct: lipid slurry typically treated as waste. Their goal was to turn it into a sustainable alternative to commercial waxes like carnauba and beeswax. But comparing these experimental materials to established wax standards wasn’t straightforward. The thermal behavior of the slurry was complex, with multiple overlapping melting events, making simple visual comparisons of differential scanning calorimetry (DSC) profiles unreliable and traditional analysis methods hard to apply.

The FDA solution

Using functional principal component analysis in JMP Pro, the team modeled DSC thermograms as continuous melting profiles, making it possible to compare experimental waxes directly to commercial standards. This approach enabled precise similarity matching to carnauba and beeswax, revealing which processing conditions produced the closest matches. By combining FDA features with machine learning, the researchers could reliably target specific wax properties, turning a bioethanol byproduct into a viable, high-value material with far less trial and error.

Adhesive science: Solving the "identical twins" problem

The challenge

Henkel engineers faced a classic materials science problem: two amine hardeners that looked virtually identical, yet behaved very differently in practice. Standard Fourier Transform Infrared (FTIR) analysis couldn’t tell them apart, even though small differences in composition had a big impact on curing performance and adhesive quality. Without a reliable way to quantify the correct ratios using routine plant measurements, ensuring consistent product performance became a serious challenge.

The FDA solution

Instead of focusing on individual FTIR peaks, Henkel engineers used JMP Pro to analyze entire spectral profiles as functional data. This made it possible to tease apart subtle but meaningful differences between the two hardeners; differences that traditional methods completely missed. The resulting model accurately quantified amine ratios and enabled reliable, plant-level quality control using standard FTIR equipment, eliminating the need for more costly techniques like Nuclear Magnetic Resonance (NMR) for routine monitoring.

From the article’s conclusion:

The benefits of combining FDA with JMP data analytics to enhance instrumental analytical capabilities were demonstrated. The FTIR-ATR spectroscopy combined with FDA was proved to be a powerful methodology to discriminate the composition of chemical mixtures with similar chemical structures.

The common thread: Why FDA succeeds where traditional methods struggle

Across these five examples, one theme is consistent: functional data analysis works because it respects how the data is actually generated.

FDA:

• Preserves the full story: Analyzing entire curves instead of isolated points.

• Reveals patterns that traditional methods miss: From early fouling signals to subtle spectral differences.

• Handles real-world complexity: Including variable batch lengths, overlapping thermal events, and noisy industrial measurements.

• Enables prediction: Allowing teams to explore new formulations, designs, or operating conditions without testing every scenario.

• Remains interpretable: Helping scientists understand why behaviors differ, not just that they do.

In each case, the breakthrough came from treating data as a process, not a spreadsheet.

From research insight to practical impact

Just as important: none of these teams needed custom code or niche academic tools. Every study used JMP Pro to apply FDA in a practical, repeatable way, often using data from instruments already in place.

Whether in pharmaceutical development, aerospace research, chemical manufacturing, or materials science, the workflow followed a familiar pattern:

• Collect continuous measurements (e.g., dissolution, DSC, FTIR, process data)
• Represent those measurements as functions
• Extract dominant patterns with FPCA
• Build predictive models tied to real outcomes
• Deploy insights for design decisions, monitoring, or quality control

The result? FDA proved effective across a wide range of settings, from academic studies to plant-floor applications, sometimes eliminating the need for additional testing or new instrumentation.

Is your data ready for FDA?

If you work with continuous measurements, chances are you’re already sitting on FDA-ready data:

• Time-varying profiles: dissolution, batch trajectories, degradation
• Temperature-dependent behavior: DSC, thermogravimetric analysis (TGA), cure profiles
• Spectroscopic signals: FTIR, near-infrared (NIR), Raman, ultraviolet–visible (UV-Vis)
• Spatial measurements: ablation, coatings, concentration gradients

If you’re reducing these rich signals to single values, such as peak heights, endpoints, or averages, you may be leaving important insight behind.