By Kunle Barike

In 2020, as Nigerian manufacturing firms contended with rising distribution costs, congested transport corridors, and increasing pressure to meet delivery timelines, Honeywell Flour Mills Plc encountered a structural limitation in its logistics planning process. 

The company’s shipment decisions spanning road, sea, and air transport had grown too complex for rule-based scheduling methods, yet too critical to be managed through intuition alone. It was at this inflection point that Sunday Oladimeji Adegoke introduced a machine learning framework that fundamentally altered how transport mode decisions were evaluated within the organization.

Rather than applying incremental automation to existing workflows, Adegoke reconceptualized the problem itself. He identified that shipment planning at scale does not conform to single-outcome decision logic; instead, each shipment may validly correspond to multiple transport modes depending on cost tolerance, delivery urgency, distance, and operational risk. 

On this basis, he framed the challenge as a multilabel classification problem, a modelling approach rarely deployed within Nigeria’s manufacturing logistics sector at the time. This reframing enabled a shift from deterministic heuristics to probabilistic, data-driven decision intelligence.

The project involved the development of a predictive system capable of recommending appropriate shipment modes for approximately 2,000 unique products, drawing on historical transport data and operational attributes. Adegoke explored and implemented several advanced multilabel learning strategies, including independent binary classifiers, classifier chain architectures, and natively multilabel algorithms such as ensemble tree models and neural networks. Each approach was rigorously evaluated using performance metrics aligned with business priorities, including accuracy, precision, recall, and operational feasibility.

Data engineering constituted a significant portion of the work. Shipment and transport records were queried and processed using cloud-based data infrastructure, enabling efficient handling of large and heterogeneous datasets. Extensive preprocessing and feature engineering were undertaken to transform raw operational data into structured inputs suitable for learning. Key variables included shipment size, product category, destination characteristics, delivery timelines, and route constraints. By embedding domain knowledge into the feature design, the models were trained to capture relationships that had previously been assessed manually by logistics planners.

The outcome was not a proof-of-concept but a deployed analytical system that materially improved decision quality. Internal evaluations at Honeywell Flour Mills indicated that the model enhanced the accuracy of transport mode selection, reduced reliance on suboptimal routing choices, and lowered planning latency. These improvements translated into measurable cost reductions and increased consistency in service delivery outcomes of particular importance in Nigeria’s cost-sensitive manufacturing environment.

Beyond immediate efficiency gains, the project demonstrated scalability. The modelling framework was designed to accommodate growth in shipment volume and complexity without proportional increases in planning overhead. This capability is especially relevant for Nigerian manufacturers seeking to expand regional distribution while maintaining operational control. By automating a previously manual, experience-driven process, the system reduced exposure to human error and improved resilience against staff turnover.

The project’s impact was formally recognized during the company’s 2020 performance appraisal cycle, in which Adegoke received an “Outstanding” rating. The appraisal cited his technical leadership, analytical depth, and ability to translate advanced machine learning techniques into tangible business value. Within Nigeria’s industrial context, such recognition underscores the increasing strategic importance of applied data science expertise.

Industry observers note that the significance of the work extends beyond logistics. Multilabel classification and predictive optimization techniques of this nature are increasingly relevant across sectors such as healthcare analytics, infrastructure planning, and financial risk management. In healthcare, for example, similar approaches are being explored globally for diagnostic decision support and outcome prediction, where multiple risk factors and intervention pathways must be evaluated concurrently.

What distinguishes Adegoke’s contribution is the combination of methodological sophistication and practical deployment within a Nigerian industrial setting. At a time when advanced artificial intelligence applications were often confined to academic research or experimental pilots, his work demonstrated that such techniques could be operationalized to address real constraints in local manufacturing and distribution systems.

As Nigeria continues to pursue industrial modernization and digital transformation, the ability to apply advanced analytics to core operational challenges has become a matter of national economic relevance. The logistics sector, in particular, plays a critical role in food security, price stability, and market access. Innovations that improve efficiency and reduce cost volatility therefore carry implications beyond individual firms.

Sunday Oladimeji Adegoke’s work at Honeywell Flour Mills illustrates how Nigerian professionals can deliver original, high-impact technological solutions that meet international analytical standards while addressing local operational realities. By reframing a complex logistics challenge through advanced machine learning and delivering measurable outcomes, his contribution represents a substantive example of indigenous innovation shaping the future of data-driven decision-making in Nigeria’s industrial economy.