Download presentation
Presentation is loading. Please wait.
Published byRosa Lamb Modified over 9 years ago
1
Effect of Neighboring Flight Patterns on a Particular Flight Presented by Venugopal Rajagopal CIS 595 Dr. Slobodan Vucetic
2
Introduction Lockheed Martin Dataset 3 months of flight data. Tables used in the Database schema are Flight, Plan, PlanPoint, Tracking About 3GB of data
3
Proposed Work Develop a Matlab display routine for: a) Visualizing flight distribution in a particular airport for a particular radius. b) Visualizing the path a particular flight takes before landing or taking off from a particular airport.
4
Proposed Work (contd.) Design a general model to predict delay using neighboring flight information. Design a model to predict delay for different airlines. Design a model to predict the difference in delta arrival and delta departure.
5
Data Preprocessing Remove data which has null value. Removing redundant data. Data Interpolation. Choosing airport hubs for data analysis.
6
Methodology Neural Networks trained to design predictors. Important attributes considered are a) Number of flights in air before departure of a particular flight b)Historical data (e.g., Average delay in departure one week before) c) Day of the week
7
Distribution of flights at a specific time Figure showing distribution of flights in a airport Figure showing flight path
8
Distribution of flights for the whole day Figure showing distribution of flights
9
Results Classifier Accuracy General model to predict delay – 70% Separate model to predict delay for different airlines: American Airlines - 71% United Airlines - 71% North West - 71%
10
Conclusion An attempt made to predict delay using neighboring flight information. Different class projects done on this dataset, it will be good if bagging or boosting techniques are used to create ensembles to increase the prediction accuracies.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.