SQL · Python · Statistics · Power BI
Completed AnalysisDelta Air Lines — Operational Performance Analysis
Independent portfolio analysis using public data. Delta Air Lines did not sponsor, endorse or supply information for this project. Delta is the strongest operational performer among the major carriers in this sample; the question is where an already reliable operation could focus attention.
Measured results
Headline measures
Delta flights analyzed
246,732
Measured
Arrivals within 15 min
81.3%
Measured
Cancellation rate
0.30%
Measured
Best major carrier, on-time
Delta (18.7% late)
Measured
Morning vs evening late rate
12.5% → 25.0%
Measured
Late-inbound aircraft → late arrival
97.9%
Measured
Overview
Project summary
- Objective
- Identify the operational conditions most associated with flight disruption at Delta, and where schedule and staffing attention would carry the greatest effect.
- Data
- Public data only: U.S. Bureau of Transportation Statistics Airline On-Time Performance records for January, June and December 2024 — 246,732 Delta flights within 2.06 million industry flights, including DOT delay-cause minutes. No internal company data and no simulated results.
- Tools
- SQL (analytical queries over 790M+ characters of raw records) · Python · Statistics · Power BI build in progress.
- Analysis
- On-time and cancellation rates benchmarked against the other major carriers; delay-cause composition by minutes; departure-hour risk curve; origin-airport performance including taxi-out time; correlation and regression of departure delay against arrival delay; conditional probability of a late arrival given a late inbound aircraft.
- Key findings
- Delta led the majors on both reliability measures — 18.7% of arrivals more than 15 minutes late versus 20.1% (United), 21.8% (Southwest) and 26.9% (American) — and cancelled 0.30% of flights against an industry range of 1.28%–3.13%. Disruption is concentrated late in the day: late arrivals double from 12.5% for 05:00–08:00 departures to 25.0% for 17:00–20:00. Controllable causes dominate the delay minutes: carrier operations (1.47M minutes) and late-arriving aircraft (0.99M) account for 73% of total delay minutes, while weather accounts for 7%. When an inbound aircraft arrives late, 97.9% of the following flight's arrivals are late, versus 11.8% otherwise. Departure delay explains arrival delay almost entirely (r = 0.965, slope ≈ 1.01) — minutes lost at the gate are not recovered in the air. Among high-volume origins, DTW (23.2% late departures), LGA (22.9%) and BOS (21.4%) run behind ATL (18.7%), which also holds the lowest taxi-out time at 16.0 minutes.
- Business impact
- Because delay does not dissipate in flight and propagates through aircraft rotations, the cost of a morning delay is paid repeatedly across every downstream leg — in misconnections, crew duty limits, recovery labor and rebooking. The 73% controllable share means most of the exposure sits inside decisions the airline owns: turn buffers, rotation design and gate resourcing, not weather.
- Recommendation
- Protect the aircraft rotation rather than the individual flight. Add turn buffer to the aircraft that feed the 17:00–20:00 band, prioritize recovery on tails already running late by midday, and target station-level turn performance at DTW, LGA and BOS, where the gap to ATL is widest under comparable volume.
Evidence
Where disruption concentrates
Late-arrival rate rises through the operating day
Share of Delta flights arriving more than 15 minutes late, by scheduled departure hour. January, June and December 2024, public BTS records.
Delay minutes are concentrated in controllable causes
Total Delta delay minutes by DOT cause category, in thousands. Carrier operations and late-arriving aircraft together account for 73% of delay minutes.
Delta leads the major carriers on arrival reliability
Share of arrivals more than 15 minutes late, carriers with more than 100,000 flights in the sample period.
Method
Representative queries
-- Late-arrival rate by scheduled departure hour
SELECT FLOOR(CRSDepTime / 100) AS dep_hour,
COUNT(*) AS flights,
ROUND(100.0 * SUM(ArrDelayMinutes > 15) / COUNT(*), 1) AS pct_late
FROM on_time_performance
WHERE Reporting_Airline = 'DL' AND Cancelled = 0
GROUP BY dep_hour
HAVING COUNT(*) > 2000
ORDER BY dep_hour;
-- Conditional probability of a late arrival, given a late inbound aircraft
SELECT CASE WHEN LateAircraftDelay > 0 THEN 'inbound late'
ELSE 'inbound on time' END AS inbound_state,
COUNT(*) AS flights,
ROUND(100.0 * SUM(ArrDelayMinutes > 15) / COUNT(*), 1) AS pct_late
FROM on_time_performance
WHERE Reporting_Airline = 'DL' AND Cancelled = 0
GROUP BY inbound_state;Business context
How reliability translates into financial performance
Operational reliability drives customer experience, missed connections, aircraft and crew utilization, recovery cost and loyalty. The measured relationships here — near-total carry-through of departure delay into arrival delay, and near-certain lateness behind a late inbound aircraft — explain why small early-day timing decisions compound into end-of-day cost.
Statistical basis
Statistical methods applied
Probability
Estimate the likelihood of an operational event.
Conditional probability
Understand how risk changes when certain conditions exist.
Mean
Establish typical performance.
Standard deviation
Measure consistency or volatility.
Regression
Explore which variables help explain an outcome.
Limitations
Scope and limitations
- — Three months of 2024 (January, June, December) were sampled to cover winter, summer and holiday operating conditions; results are not a full-year measure.
- — Public on-time data reports delay minutes by category; the categories are reported attributions, not established causes.
- — Delay-cause minutes are reported only for flights delayed 15 minutes or more, so cause shares describe delayed flights rather than all flights.
- — No internal cost data exists publicly, so financial impact is described as direction and mechanism, never estimated in dollars.