Determine Drivers of Airline Flight Delays

Context

You are analyzing a flight-level dataset to identify which factors most impact delays. Assume you have one row per flight with columns such as:

• delay_min (arrival delay in minutes; can be 0+ and skewed)
• delayed_15 (binary: delay_min ≥ 15)
• carrier, origin, destination, aircraft_type
• scheduled_departure_hour, day_of_week, month (seasonality), holiday
• route_distance, precipitation, wind, visibility (origin/destination weather)
• flight_date (for time-aware validation)

Task

Propose a statistical approach to determine which factors most impact delays. Specifically:

1. Choose appropriate outcome(s) and justify them.
2. Specify models and hypothesis tests you would use to quantify factor impacts.
3. Detail how you would validate model assumptions and guard against common pitfalls (e.g., seasonality, heteroskedasticity).
4. Explain how you would interpret results and report uncertainty.

Hints: Consider regression, ANOVA/Type II/III tests, confidence intervals, seasonality modeling, heteroskedasticity checks, and time-aware validation.