Date of Award
As the Age of Information has evolved over the last several decades, the demand for technology which stores, analyzes, and utilizes data has increased substantially. For countless industries such as the medical, retail, and aircraft industries, such technology is crucial to their operation. This project proposes a hybrid machine learning model consisting of Decision Trees and Neural Networks which is able to classify data of varying volume and variety effectively and efficiently. The model’s structure consists of a decision tree with each node of the tree containing a neural network trained to classify a specific category of the output using binary classification. To validate the model’s efficacy, it is tested by applying it to a dataset consisting of the Federal Aviation Administration’s (FAA’s) Boeing 737 maintenance data, consisting of 137,236 unique records, each comprised of 72 variables, in a predictive maintenance setting. The predictive maintenance is performed by classifying the Discrepancy variable, a free-text descriptor of the maintenance issue faced by the aircraft, by first determining if the issue occurred during scheduled maintenance or not, and subsequently breaking down the nature of the incident into more specific categories. Results indicate that this hybrid model is able to classify incidents with high accuracy and precision. Additionally, the model is able to identify the most significant inputs involved in classification allowing for increased model performance. This both demonstrates the model’s applicability to real-world scenarios and showcases the benefits of combining Decision Trees and Neural Networks in a hybrid structure rather than using them individually.
Carson, Jarrod, "A Hybrid Decision Tree - Neural Network (DT-NN) Model for Predictive Maintenance Applications in Aircraft" (2021). Undergraduate Theses . 11.