1-3 of 3 results
-
Predicting General Aviation Accidents Using Machine Learning Algorithms
PI Bradley Baugh
CO-I Bruce Conway
Aviation safety management is implemented through reactive, proactive, and predictive methodologies. Unlike reactive and proactive safety, predictive safety can predict the next accident and enable prevention before an actual occurrence. The study outlined here promotes predictive safety management through machine learning technologies using large amounts of data to facilitate predictive modeling.
The study addresses efforts to reduce General Aviation accidents, an effort that was renewed in earnest with the Federal Aviation Administration’s 1998 Safer Skies Initiative. Over the past 22 years, the General Aviation fatality rate has decreased. However, accidents still happen, and there is some evidence showing the number of accidents, representing hazard exposure, is increasing. The accident data suggest that the aviation community still has more to learn about the variables involved in an accident sequence.
The purpose of the study was to conduct an exploratory data-driven examination of General Aviation accidents in the United States from January 1, 1998, to December 31, 2018, using machine learning and data mining techniques. The goal was to determine what model best predicts fatal and severe injury aviation accidents and further, what variables were most important in the prediction model.
The study sample comprised 26,387 fixed-wing general aviation accidents accessed through the publicly accessible National Transportation Safety Board Aviation Accident Database and Synopses archive. Using a mixed-methods approach, the study employed both unstructured narrative text and structured tabular data within the predictive modeling. First, the accident narratives were culled using text mining algorithms to develop text-based quantitative variables. Next, data mining algorithms were used to develop models based on both text- and data-based variables derived from the accident reports.
Five types of machine learning models were created using SAS® Enterprise Miner™, including the Decision Tree, Gradient Boosting, Logistic Regression, Neural Network, and Random Forest. Additionally, three broad sets of variables were used in modeling, including text-only, data-only, and a combination of text and data variables. Three models, Logistic Regression (text-only variables), Random Forest (text-only variables), and Gradient Boosting (text and data variables), emerged with a similar prediction capability. The top six variables within the models were all text-based covering Medical, Slow-flight and stalls, Flight control, IMC flight, Weather factors, and Flight hours topics. The Logistic Regression (Text) model was selected as the champion model: Misclassification Rate = 0.098, ROC Index = 0.945, and Cumulative Lift = 3.46.
The results of the study provide insights to the entire General Aviation community, including government, industry, flight training, and the operational pilot. Specific recommendations include the following areas: 1) improve the quality and usefulness of accident reports for machine learning applications, 2) investigate ways to capture and publish more open-source flight data for use in safety modeling, 3) invest in additional medical education and find ways to address impairing medications and high risk medical conditions, 4) renew efforts on improving flight skills and combatting decision-based errors, 5) emphasize the importance of weather briefings, pre-flight planning, and weather-based risk management, and 6) create an aviation-specific corpus for text mining to improve text analysis and transformation.
Categories: Graduate
-
An Exploratory Study of General Aviation Visual to Instrument Meteorological Condition Contextual Factors
PI James Hartman
CO-I Mark Friend
The purpose of this dissertation was to bridge the existing literature gap of outdated contextual factor (CF) research through examination and determination of current General Aviation (GA) Title 14 Code of Federal Regulations (CFR) Part 91 visual flight rules (VFR)-into-instrument meteorological condition (IMC) contextual factors. Contextual factors are a multifaceted arrangement of pertinent events or occurrences contributing to pilot accidents in weather-related decision-making errors.
A total of 46 contextual factors were identified and examined from the reviewed research literature. The study examined and determined the presence of the 46 contextual factors, frequencies, and manifestations in the GA VFR-into-IMC Aviation Accident Reports (AARs) archived in the National Transportation Safety Board (NTSB) online safety database. Significant relationships were identified among the contextual factors and pilot age, flight experience, weather, flight conditions, time of day, and certification level using point biserial and phi correlations. Contextual factor significant effects on the crash distance from departure and crash distance from the planned destination were revealed using multiple regression. A qualitative methodology was used on secondary data. Three subject matter experts (SMEs) for the main study analyzed a sample of 85 accidents for the presence of the 46 contextual factors. Raters then reported the presence of the contextual factors and provided opinions on how the contextual factors were manifested. Qualitative analysis revealed the presence of 37 out of 46 contextual factors. Highest frequency factors included number of passengers on board (CF29), accident time of day (CF1), crash distance from the planned destination (CF15), not filing of a flight plan (CF21), and underestimating risk (CF43). Raters described numerous manifestations of the contextual factors including 62% of the accident flights had passengers on board the aircraft (CF29). Quantitative analysis discovered several significantly weak to moderate relationships among pilot age, flight experience, weather, flight conditions, time of day, certification level, and the contextual factors. Several contextual factors had significant effects on the crash distance from departure and crash distance from the planned destination. Findings indicated the contextual factors were extensive in GA accidents. Additional research should focus on all flight domains, including further study of GA Part 91 VFR-into-IMC accidents. It is recommended the GA Part 91 pilot community be trained on the contextual factors assessed.
Categories: Graduate
-
Assessing If Motivation Impacts General Aviation Pilots’ Persistence in Varying Weather Conditions
PI Sabrina Woods
CO-I Scott Winter
Continued flight under visual flight rules into instrument meteorological conditions is the predominant cause for fatal accidents by percentage, for general aviation aircraft operations. It is possible that a pilot’s motivation or reason for flying will override other safer, more logical courses of action when a hazard presents itself. The decision appears to stem from a willingness to persist in a course of action despite factors that indicate an alternate and safer course is warranted. This research addresses what is currently presumed about the decision to continue flying under visual flight rules into instrument conditions and marries those ideas with the extensive studies on how theoretically affects the decision-making process.
The research used a quantitative factorial experimental design and explored what bearing, if any, does type of motivation, or meteorological condition, or the interaction of the two have on a pilot’s willingness to persist in visual flight rule into instrument meteorological conditions. The researcher applied fundamental motivation theory and aviation regulation in the development of scenarios that were used to assess a pilot’s willingness to persist in unsafe weather conditions, and to determine what role motivation and the weather conditions might have played in that decision. A 3x3 factorial design was followed, and the method of analysis was a two-way mixed analysis of variance.
The independent variable meteorological condition indicated a significant effect on the dependent variable willingness to persist, and the independent variable motivation did not indicate a significant effect. The interaction between meteorological condition and motivation resulted in a significant effect on the dependent variable, particularly in the marginal weather condition, although with a low effect size. This result suggests that those who are motivated to fly for a specific reason or reasons might be more willing to persist over those who have no real reason to be flying. A recommendation for future research is that the experiment be replicated in a direct observation experimental design in either a full or partial motion simulator.
Further defining how motivation and meteorological conditions influence aeronautical decision-making can change the way aviation safety advocates, academics, regulators, and industry approach the issue. The results of this research could help determine what part of aeronautical decision making is objective and what is more subject to a person’s base desires.
Categories: Graduate
1-3 of 3 results