Nigerian engineer advances transportation workforce planning with cutting-edge AI model

Amid a rising concern for effective workforce planning in state transportation agencies, Adedolapo Ogungbire, a leading expert in civil engineering and machine learning, has expressed optimism about the potential of machine learning to address workforce shortages in the construction sector.

Ogungbire’s research focuses on developing machine learning models to forecast person-hour requirements for construction engineering positions, which are critical for managing transportation projects as lane miles increase.

His recent thesis explored innovative forecasting solutions that could transform workforce planning, allowing agencies to better allocate resources and manage personnel.

The expert, who interacted with newsmen during a seminar on transportation innovation, noted that a critical component of effective workforce planning is accurate forecasting at both the agency and project levels. He emphasized that traditional methods of estimating workforce needs often lack precision, which has increasingly become an issue in states with rapidly expanding infrastructure. Ogungbire’s research suggests that artificial intelligence, particularly machine learning, can make these forecasts more reliable and adaptable to the fluctuating demands of transportation agencies.

He explained how his study used data from the Arkansas Department of Transportation (ARDOT) between 2012 and 2021 to build machine learning models that estimate workforce requirements. His models, ranging from linear regressions to neural networks, demonstrated the potential to predict workforce needs for both short-term and long-term projects. “By leveraging historical data, we can now make more informed decisions regarding the number of hours and engineers needed for specific projects, reducing wasted resources and improving project efficiency,” Ogungbire shared.

At the project level, his team developed machine learning models that use regressors from linear to advanced neural networks, aiming to provide high-accuracy workforce estimates for individual projects. Among the machine learning methods tested, the random forest regressor—a type of ensemble model that uses decision trees and bagging techniques—proved especially effective, achieving an impressive R-squared value of 0.91. This model’s high accuracy demonstrates that machine learning tools can reliably estimate staffing needs for infrastructure projects.

On an agency-wide scale, Ogungbire’s research applied both classic time series methods and neural networks to forecast monthly person-hour requirements. His most effective tool for this type of forecasting was a one-dimensional convolutional neural network model. This model recorded an average root mean square error (RMSE) of 4,500 person-hours monthly for short-range forecasting, indicating a high level of accuracy for immediate workforce planning. The model performed well even over extended periods, with an RMSE of 5,000 person-hours in long-range forecasts.

These findings highlight the practical advantages machine learning offers to agencies such as ARDOT, as they prepare for future projects and navigate workforce shortages.

Ogungbire pointed out that while workforce planning is a long-standing challenge for transportation agencies, his research represents a substantial step forward in addressing it, particularly at a time when many state agencies are experiencing construction delays due to limited personnel.

Ogungbire’s current role as a Graduate Research Assistant at the University of Arkansas’ ZeRo Lab has allowed him to expand his expertise. He leads analytics and dashboard deployment for SMILIES, a pilot bikeshare project funded by the National Science Foundation, demonstrating his capacity to manage both research and practical applications in transportation planning.

In addition to his academic contributions, Ogungbire has served as a Transportation Planning Consultant at Leedigital Design Consults since 2020.

His work has involved developing state-of-the-art transportation plans across Oyo State, Nigeria, including projects aimed at easing high-traffic congestion in Ibadan.

He also spearheaded the development of an interactive, open-source transportation planning software tailored to rural communities, showcasing his commitment to creating accessible solutions.

Prior to his consulting work, Ogungbire gained hands-on experience as a Graduate Engineer at the Ekiti State Ministry of Works and Transport, where he supervised the Ado-Iyin Road construction. His role included forecasting manpower requirements and digitizing construction data, further informing his perspective on workforce planning and forecasting challenges.

Ogungbire’s technical background is extensive, with skills in Python, R, SQL, Java, Stata, and MATLAB, among others. His proficiency in data visualization tools like Tableau and Power BI has been instrumental in his research, allowing him to present complex workforce projections in an accessible format for agencies to use in decision-making.

His work has not gone unrecognized. Ogungbire received the prestigious Tau Beta Pi Graduate Fellowship in 2024 and was named a Distinguished Doctoral Fellow in 2023. His other accolades include the Carolyn Clark Langston & Dr. Harold D. Langston Endowment for Technical Outreach in 2022 and the Graduate Student of the Year award for his master’s work at the University of Arkansas.

Ogungbire’s accomplishments reflect a dedication to bridging academic research with real-world applications in workforce forecasting. By combining his engineering background with machine learning, he aims to address workforce planning issues that have historically impacted transportation agencies, particularly as infrastructure demands rise.

Looking to the future, Ogungbire envisions broader applications of his research beyond the Arkansas Department of Transportation. “This technology could be adapted to other industries facing similar workforce challenges, enabling data-driven solutions that optimize personnel resources effectively and efficiently,” he said.

As Ogungbire’s research continues to evolve, his work serves as an example of how machine learning can be used to solve complex issues in workforce planning. His contributions have the potential to make significant improvements in transportation agencies’ operational strategies, underscoring the value of advanced data analysis in the field of civil engineering.