Contents 1. Introduction 3 2. Executive Summary 5 3. Data Preprocessing 6 4. Analysis 8 4.1 Data Partition 8 4.2 Stat Explore 10 4.3 Clustering & Segmentation: 12 4.4 Decision Tree 24 4.5 Interactive Decision Tree 28 4.6 Gradient Boosting 33 4.7 Linear Regression 35 4.8 Neural Network 38 4.9 Compare Models 40 4.10 Score New Data 42 4.11 Logistic Regression 44 5. Conclusion 49

1. Introduction
Given the complexity and the large extent of the interdependencies between airports, aircraft, passengers, airlines, control centers, etc. of the national aviation system, flight delays occur frequently. Recently, the OAG Flight Status database reported that over 4.6 million flights arrived more than 15 minutes late at their destination; a conservative average of 80 passengers per flight equates to about 368 million passengers being inconvenienced. Inspired by this, our term paper predicts the delay in flights of some of the players in the U.S. Airlines Industry and the impact of the flight delays to improve their performance (July 2014 to June 2015).

DATA SOURCE
The secondary data source we will use is available under Airline On-time Statistics on the TranStats website of the Bureau of Transportation Statistics as our source of data. We extracted information for specific carriers from all major airports during a particular time period. For the purpose of our analysis, we restricted the scope of our study to focus on the arrivals in time period between July 2014 and June 2015, rendering it more relevant. We have included the current top 14(according to revenue) flight carrier companies in United States– American Airlines, Delta Airlines, Southwest Airlines, United Airlines, JetBlue Airways, Alaska Airlines, EVA Airlines, Frontier Airlines, Air-Tran Airlines, Hawaiian Airlines, American Eagle, SkyWest Airlines,