Application of Data Analytics and Machine Learning Methods to Enhance Decision-Making in Right-Of-Way Acquisition Process and Transportation Asset Management
Author(s)
Chung, Frederick B.
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
This research explores implementation of data analytics and machine learning techniques to enhance two crucial project management tasks, Right-Of-Way (ROW) acquisition process and transportation asset management. The first chapter of this research contributes to the state of knowledge in estimating ROW acquisition timeline through developing a novel machine learning model to accurately estimate ROW acquisition timelines, and identifying drivers (i.e., risk factors) of ROW acquisition durations. The forecasting model developed in this research achieves a high accuracy to predict ROW durations by explaining 74% of the variance in ROW acquisition durations using project features. Moreover, number of parcels, average cost estimation per parcel, length of projects, number of condemnations, number of relocations, and type of work are found to be influential factors as drivers of ROW acquisition duration. The second chapter of this research contributes to the body of knowledge in improving the prediction of pavement condition by developing machine learning models and implementing ensemble methods to enhance predictive performance. This research focuses on developing five machine learning classification models, Random Forest, Gradient Boosting, Support Vector Machine, K-Nearest Neighbor, and Artificial Neural Network, to predict pavement condition levels. To enhance prediction performance, ensemble methods, including voting and stacking, are integrated. Voting ensemble model constructed with the two best-performing individual classification models reaches the highest accuracy rate at 83%. Although the performance of individual models fluctuates, ensemble models consistently produce top-tier performance irrespective of the data sampling variations. Therefore, ensemble methods are recommended in developing pavement condition prediction models to enhance accuracy and achieve more consistent quality of predictions. The findings of this research will provide transportation agencies with insights on how to improve practices in scheduling ROW acquisition process and improving pavement condition forecasting practices to enhance their maintenance planning and cost savings.
Sponsor
Date
2024-06-05
Extent
Resource Type
Text
Resource Subtype
Dissertation