Business PhD Spotlight: Erfan Rafieikia

PhD candidate Erfan Rafieikia’s journey into advanced analytics was shaped by a blend of family values and the digital revolution. Raised in an environment that prioritized a growth mindset and a relentless passion for learning, he was naturally drawn to complex problem-solving. 

During his undergraduate years, Rafieikia witnessed the explosive rise of the e-commerce era. Seeing how global titans like Amazon and Uber leveraged data-driven decision-making and machine learning as their "DNA", this inspired him to look beyond the surface of business operations.

To bridge the gap between theory and practice, Rafieikia transitioned into the industry sector to see these data-centric systems in action. This experience deepened his curiosity.

"Working in the industry sector allowed me to see the intricate nuances and the vast knowledge required to steer these systems," Rafieikia explains. "I realized that to truly master these complexities, I needed to go deeper. That realization was the catalyst for my PhD journey — a quest for a profound understanding of how data and logic intersect."

Research focus: integration of optimization and intelligence

By developing learning-augmented optimization methods, Rafieikia aims to improve large-scale decision-making under uncertainty. For instance, this research helps businesses determine how to price products without knowing exactly what future demand will look like.

Rafieikia is currently working on three research projects in this broad area.

Accelerating large-scale optimization

In one research project, Rafieikia develops a new variant of the Benders Decomposition (BD) algorithm — an established method for solving large-scale optimization problems — that integrates machine learning into the solution process. Classic BD methods can be computationally slow, limiting their practicality in fast-moving industries such as retail, transportation, and healthcare, where timely decisions are critical.

By using ML to predict key structural information within subproblems, his approach significantly accelerates solution times. This enables organizations to solve complex, large-scale operational problems more efficiently.

Delivery integration in markdown platforms

Globally, nearly one-third of all food produced is wasted, while rising food prices make access to affordable food increasingly difficult for many households. Reducing waste and expanding food accessibility are therefore two sides of the same global challenge.

To tackle this dual problem directly, Rafieikia’s second research project studies markdown platforms — digital marketplaces that offer near-expiration food at discounted prices — as a mechanism to address both issues. His work focuses on improving these systems by integrating delivery services, which can expand access to affordable food and further reduce waste.

A key complexity arises from the strategic interaction between the platform and retailers. The platform sets delivery fees, while retailers determine discount prices — both of which influence customers’ perceived value and purchasing decisions. To capture this interaction, Rafieikia models the system as a Stackelberg game, where the platform acts as a “leader” and retailers as “followers”. This leader–follower structure is formulated as a bilevel optimization problem, allowing the research to explicitly analyze how pricing decisions at one level shape outcomes at the other.

A central challenge is determining how delivery pricing should be structured so that platforms remain financially sustainable while increasing accessibility for consumers. By developing data-driven pricing models within this framework, the research aims to support environmental sustainability while improving affordability.

Data-driven discount decisions

In a third research project, Rafieikia focuses on discount pricing itself. Pricing near-expiration products presents a fundamental challenge: setting prices too low reduces profitability, while setting them too high can result in unsold inventory and food waste.

This project investigates how contextual information — such as product characteristics, timing, and remaining shelf life — can be leveraged to guide better pricing decisions. By designing data-driven discount policies, the research helps sellers balance profitability with waste reduction, supporting smarter and more sustainable retail operations.

Developing these kinds of sophisticated, data-driven solutions requires an academic environment at the forefront of technological innovation. When it came time to choose a business doctoral program, the University of Alberta was the clear choice. "Alberta is one of the primary AI pillars of Canada," says Rafieikia. "The province has a rich ecosystem for innovation, and the University of Alberta is at the heart of that."

Under the guidance of his supervisor, Dr. Borzou Rostami, Rafieikia is an active member of the AI Centre for Decision Analytics (AI4DA) within the Alberta School of Business. This research center provides a unique collaborative platform where Rafieikia can apply his business research to diverse, high-impact sectors, including government initiatives, healthcare systems, and retail operations. Working within AI4DA has provided him with the ideal environment to transform theoretical breakthroughs into actionable solutions that address the specific needs of these varied industries.