Applying Machine Learning Technique to Improve Cost Estimating Process for Tennessee’s Transportation Projects
Abstract
According to the 2021 America’s Infrastructure Report Card by the American Society of Civil Engineers, 43% of U.S. roadways are in poor or mediocre condition, costing drivers over $1,000 per year in wasted time and fuel. A key challenge is the inaccurate estimation of road repair and maintenance costs, which hinders efficient fund allocation. While extensive research exists on cost prediction for new construction, infrastructure improvement projects, such as bridge repairs and road rehabilitations, receive less attention despite their unique complexities. This research addresses that challenge by developing a Cost Prediction Tool using Excel. The tool is specifically designed to assist Tennessee Department of Transportation (TDOT) engineers in overcoming difficulties with cost estimation during the planning phase. An in-depth analysis was done on 11 years of road and bridge project data across different regions of Tennessee, and various factors like project length, route, work type, and location were considered for developing the tool. Machine learning models were developed to predict costs at the state, regional, and county levels for projects such as bridge repairs, safety improvements, and legislative programs. These models were validated with new data, and the validation results demonstrated strong performance in early project phases. A spreadsheet-based tool was then created to make cost estimation quick and easy. This tool will be particularly valuable in emergencies, such as natural disasters like Hurricane Helena, by helping TDOT to fast-track the approval and funding for urgent repairs, ensuring a faster response and resource allocation.
Start Time
16-4-2025 1:30 PM
End Time
16-4-2025 2:30 PM
Room Number
303
Presentation Type
Oral Presentation
Presentation Subtype
Grad/Comp Orals
Presentation Category
Science, Technology and Engineering
Faculty Mentor
K. Joseph Shrestha
Applying Machine Learning Technique to Improve Cost Estimating Process for Tennessee’s Transportation Projects
303
According to the 2021 America’s Infrastructure Report Card by the American Society of Civil Engineers, 43% of U.S. roadways are in poor or mediocre condition, costing drivers over $1,000 per year in wasted time and fuel. A key challenge is the inaccurate estimation of road repair and maintenance costs, which hinders efficient fund allocation. While extensive research exists on cost prediction for new construction, infrastructure improvement projects, such as bridge repairs and road rehabilitations, receive less attention despite their unique complexities. This research addresses that challenge by developing a Cost Prediction Tool using Excel. The tool is specifically designed to assist Tennessee Department of Transportation (TDOT) engineers in overcoming difficulties with cost estimation during the planning phase. An in-depth analysis was done on 11 years of road and bridge project data across different regions of Tennessee, and various factors like project length, route, work type, and location were considered for developing the tool. Machine learning models were developed to predict costs at the state, regional, and county levels for projects such as bridge repairs, safety improvements, and legislative programs. These models were validated with new data, and the validation results demonstrated strong performance in early project phases. A spreadsheet-based tool was then created to make cost estimation quick and easy. This tool will be particularly valuable in emergencies, such as natural disasters like Hurricane Helena, by helping TDOT to fast-track the approval and funding for urgent repairs, ensuring a faster response and resource allocation.