avihayamor/US_flight-delay-2024

U.S. Flight Delay Dataset — 2024

Presentation Video

For quick reference, the key findings are outlined below.

Dataset Overview

• Source: Kaggle — Flight Data 2024
• Size: 100,000 rows sampled, 35 columns
• Target Variable: arr_delay — Arrival delay in minutes (positive = late, negative = early)
• Goal: Predict flight arrival delay in minutes based on scheduled timing, departure performance, and operational metrics.

---

Data Cleaning Decisions

• Removed cancelled flights (cancelled == 0) — they have no arrival time
• Removed rows with missing arr_delay values
• Removed extreme outliers: delays below -100 or above +300 minutes (less than 0.3% of data)
• No duplicate rows were found

---

Key Findings

• Flights that depart late usually arrive late (0.90 correlation)
• Delays build throughout the day and peak around evening hours
• Florida airports showed the highest average delays in the sample
• Distance and airtime had almost no relationship with delays
• Regional and premium carriers performed better on average

---

Target Variable: arr_delay

The target variable is arr_delay (Arrival Delay), which represents the difference in minutes between the scheduled and actual arrival time of a flight.

• Positive values → the flight arrived late
• Negative values → the flight arrived early
• Zero → the flight arrived exactly on time

Since arr_delay is a numeric value measured in minutes, predicting it is a regression task..

---

Research Questions & Insights

Q1: Which airlines have the highest average arrival delay?

Airlines operate very differently in terms of scheduling and efficiency. By comparing average arrival delay per carrier, we can identify the most and least punctual airlines.

Insight: B6 (JetBlue) and NK (Spirit) have the highest average delays (~9–6 minutes late). YX (Republic Airways - American Eagle, Delta Connection, United Express) 9E (regional fully owned by Delta), and DL (Delta) consistently arrive ahead of schedule. Budget and low-cost carriers tend to run later than regional ones.

---

Q2: Does the time of day affect arrival delays?

Flights later in the day may accumulate delays due to earlier flights running behind — the "delay snowball" effect.

Insight: Early morning flights (5–6am) arrive significantly ahead of schedule due to low airport congestion. Delays build up through the day, peaking around 7pm. This confirms that delays cascade as the day progresses.

---

Q3: Which departure airports have the highest average arrival delay?

Insight: Florida airports (MIA, MCO) are the most delay-prone, likely due to heavy tourist traffic and frequent thunderstorms. MSP, BOS, and ATL tend to perform better on average.

🗺️ Open the interactive map

---

Q4: How do numerical features correlate with arrival delay?

A correlation heatmap shows how strongly each variable relates to the others, helping identify the best predictors of arr_delay.

Insight: dep_delay has a 0.90 correlation with arr_delay — by far the strongest predictor. If a flight departs late, it almost certainly arrives late. carrier_delay (0.61) and late_aircraft_delay (0.63) also contribute significantly. Interestingly, distance and air_time have almost zero correlation with delays — how far or long a flight is doesn't affect whether it's delayed.

---

Author

Avihay Amor
B.A. Economics & Entrepreneurship
Reichman University
2026

Files in this Dataset

• flight_data_2024_sample.csv — cleaned sample of 100,000 flights
• Assignment_1_EDA_&_Dataset_Avihay_Amor.ipynb — full EDA notebook
• q1_airlines.png — bar chart: average delay by airline
• q2_time_of_day.png — bar chart: average delay by hour
• q3_map.html — interactive map: delay by departure airport
• q4_heatmap.png — correlation heatmap
• Data_Science_Vid.MP4 — presentation video