Nigerian PhD Student Develops Explainable AI Model to Transform E-Commerce Forecasting

In an era where artificial intelligence powers global commerce but often feels like an opaque black box, one Nigerian scholar is challenging that narrative with groundbreaking research designed to make machine learning models more transparent, practical, and trustworthy.

Deborah Okoli, a Nigerian and current PhD candidate in Applied Mathematics at Mississippi State University, has developed a machine learning framework tailored for e-commerce that is both explainable and actionable. Her research, which she describes as “lag-aware machine learning,” addresses one of the most overlooked but critical challenges in digital markets: not just what drives growth, but when those factors take effect.

“My goal is simple — to build machine learning models that leaders can question and trust,” Okoli told in an interview. “Models that don’t just give answers, but explain how they got there.”

Beyond the Black Box

While AI-driven predictions are increasingly used by e-commerce giants, most decision-makers remain skeptical due to the “black box” nature of algorithms. Okoli’s model directly addresses this challenge by integrating time-shifted featuresthat map not only the strength but also the timing of economic influences such as:

• Labour productivity

• R&D investments

• Sales and revenue trends

• Employment shifts

• Capital expenditure

Instead of merely correlating data, Okoli’s method explains whether a rise in productivity leads to immediate sales growth or if, for instance, R&D spending impacts revenues months down the line.

Her approach combines regularised regression models, tree-based algorithms, and explainability tools like feature attribution and partial dependence plots. The result is a system that tells business leaders why predictions were made, not just what the numbers say.

“Forecasts should come with seatbelts,” Okoli explained. “If one new data point could drastically change the outcome, decision-makers deserve to know that upfront.”

Rigor Meets Real-World Application

Unlike typical academic research that lives in theory, Okoli’s models are stress-tested against real-world conditions. She uses rolling-window cross-validation, stability checks across time lags, and residual diagnostics to ensure robustness. Each model is also benchmarked against traditional economic baselines to confirm its added value.

The outcome: clear, reliable forecasts with confidence ranges that businesses can actually use. For example, her work has shown that while productivity spikes may trigger sales increases within the next quarter, R&D’s impact often unfolds over longer periods — critical insights for inventory management, logistics planning, and digital strategy.

From Nigeria to Global AI Research

Okoli’s journey reflects both intellectual brilliance and resilience. Born and raised in Nigeria, she graduated with First-Class Honours in Industrial Mathematics from Covenant University, where she emerged as the top student in her department.

She later pursued graduate studies at Tennessee Tech University, before transferring to Mississippi State University for her PhD. In addition, she holds a Master’s in Education Research from the University of Hull, UK.

At Mississippi State, she works under the mentorship of Professor Kim Seongjai (Mathematical Sciences) and Professor Jason Shin (College of Business), developing forecasting templates that organisations worldwide can adapt to their own data.

A Vision for Accessible AI

Looking ahead, Okoli hopes her work can democratize AI in ways that empower both large corporations and small businesses.

“I want to create plain-language forecasting tools for non-technical users,” she said. “AI templates that small businesses and agencies can adopt, and open collaborations that raise global standards for clarity and accountability in machine learning.”

Her mission is clear: to make artificial intelligence less intimidating, more transparent, and ultimately, more useful.

As e-commerce continues to expand at breakneck speed — with projections that global online sales could surpass $8 trillion by 2030 — innovations like Okoli’s could prove pivotal in shaping how businesses prepare for the future.

Why It Matters

Okoli’s research arrives at a time when policymakers, investors, and entrepreneurs increasingly demand explainable AI (XAI) solutions. While predictive power is essential, so is trust — especially in markets where billions of dollars ride on forecasting accuracy.

Her work doesn’t just advance machine learning; it redefines what responsible AI can look like in the 21st century.

In her own words:
“When machine learning becomes easy to understand, it becomes trustworthy. And when it’s trustworthy, it becomes useful.”

About the Author:
Samuel  is a contributing writer for 9jafinds, covering innovation, technology, and African excellence on the global stage.