Documents
Download the Final Report below.
finalreport.pdf |
Download the C++ files below that were used to filter the data into usable datasets.
- filtertop5.cpp was used to filter the whole dataset into flights leaving and landing from the top five busiest airports in the United States.
- filterregressionandclass.cpp was used to filter the top5-all dataset into training and test sets for regression models and three-class classifiers.
- filtersplitdata.cpp was used to filter the top5-all dataset into a training and test set that excluded flights that were late but by less than an hour.
|
|
|
Download the .csv data files below.
- train_and_test_data contains the datasets used for training and testing the classifiers.
- original_data contains the raw data for all ~48,000,000 domestic flights in the USA between 2010 and 2017 and the raw data for the ~1,000,000 of those flights that both left from and landed at one of the top five busiest airports in the United States.
|
Click here to download original_data.zip (download from dropbox.com)
|
Download the weka results below.
- wekamodels contains the models that were trained on my data. They can be downloaded, loaded into weka, and used to produce results on more flight data provided it uses the same attributes as the model
- resultbuffers contains the text buffers produced by weka after each classifier was run, including advanced result data
|
|