Smarter Infrastructure: Using Technology to Create Smarter Cities
Grade Level at Time of Presentation
Senior
Major
Business Data Analytics
Minor
Computer Information Systems
Institution
Western Kentucky University
KY House District #
25
KY Senate District #
10
Faculty Advisor/ Mentor
Dr. Lily Popova Zhuhadar
Department
Computer Information Systems
Abstract
I believe it’s essential to take advantage of data to provide governments and organizations with smarter and innovative solutions to cities’ problems. In this research, I examine one of the most critical issues that concern cities and communities throughout the world—the deterioration of transportation infrastructure. To this end, I utilize a dataset that contains attributes of 493 bridges that were recently assessed in cities for safety and repair needs. One of the attributes, in this dataset, classifies each bridge (in order of urgency) into four categories (monitor, schedule assessment, schedule repair/replacement, and immediate repair/replacement.) Accordingly, I train a machine learning algorithm with the historical dataset of these 493 bridges to create a predictive model. With this context, I examine a range of machine learning algorithms (Discriminant Linear Analysis, k-Nearest Neighbors, and Naïve Bayes), and compare the results. Finally, the best fitting model will be selected. The chosen model is capable of predicting the status (category) of a new uncategorized bridge.
Smarter Infrastructure: Using Technology to Create Smarter Cities
I believe it’s essential to take advantage of data to provide governments and organizations with smarter and innovative solutions to cities’ problems. In this research, I examine one of the most critical issues that concern cities and communities throughout the world—the deterioration of transportation infrastructure. To this end, I utilize a dataset that contains attributes of 493 bridges that were recently assessed in cities for safety and repair needs. One of the attributes, in this dataset, classifies each bridge (in order of urgency) into four categories (monitor, schedule assessment, schedule repair/replacement, and immediate repair/replacement.) Accordingly, I train a machine learning algorithm with the historical dataset of these 493 bridges to create a predictive model. With this context, I examine a range of machine learning algorithms (Discriminant Linear Analysis, k-Nearest Neighbors, and Naïve Bayes), and compare the results. Finally, the best fitting model will be selected. The chosen model is capable of predicting the status (category) of a new uncategorized bridge.