streamlit run dashboards/app.pyOpen http://localhost:8501 in browser.
"AeroInsight is an AI-powered airline intelligence platform built on federal BTS data — 351,000 route records across 44 carriers and 17,000+ routes."
Show: KPI cards (carriers, routes, departures, aircraft types) and the 4 key findings cards.
"The platform answers 4 critical business questions. Let me walk through each."
Click: "Competitive Intelligence" in sidebar.
"First — why does Southwest dominate domestic? They have the lowest CASM at 8.7 cents, 22% below the major carrier median. Their single-fleet strategy with 737s and highest gauge at 175 seats per departure spreads fixed costs effectively."
Show: Market share bar chart, then cost vs market share scatter plot.
"35% of US domestic routes are monopolies. Delta controls 68% of Atlanta departures."
Click: "Aircraft ROI" in sidebar.
"Which aircraft has the best economics? The B737 MAX 8 leads at 7.3 cents CASM."
Show: CASM rankings bar chart.
"A common question: Boeing 737 or Airbus A320? Our analysis shows they're virtually identical — 9.70 cents versus 9.69 cents. Fleet decisions should be driven by delivery timelines and maintenance commonality, not cost differences."
Show: 737 vs A320 comparison section.
"We also built an ML recommender — XGBoost with 93.7% accuracy — that predicts the optimal aircraft category for any route."
Click: "Cost Savings" in sidebar. Select United Airlines.
"Where can airlines save money? For United, we identified that replacing their A320 and B737-800 fleet with B737 MAX 9 saves nearly 6 cents per seat-mile."
Show: Replacement suggestions chart.
"At their Chicago O'Hare hub, we analyzed 92 routes and found significant redeployment potential. The what-if tool lets planners model specific fleet swaps."
Show: Hub optimization section.
Click: "Forecasting" in sidebar. Select American Airlines.
"What will costs look like next quarter? With only 3 quarters of data, we can't use Prophet or LSTM. Instead, we use cross-carrier panel trend extrapolation — leveraging patterns across 40+ carriers."
Show: Fuel price forecast chart with confidence intervals.
"American Airlines' Q4 fuel price is forecast at $2.36 per gallon, up 6% from Q3, with total fuel costs around $806 million."
Show: Budget planning scenarios.
"JetBlue is flagged as the highest cost risk — both fuel and CASM trending up."
Click: "Data Explorer" in sidebar.
"Every analysis is backed by transparent data. Users can browse, filter, and download the underlying dataset — 351,000 rows across 76 columns."
Show: Column selector, filter by carrier, download button.
"AeroInsight was built in 9 phases: data pipeline, feature engineering, EDA, then 4 analytical modules, dashboard integration, and testing. The codebase includes 62 automated tests, comprehensive documentation, and handles known data limitations like the P52 scope mismatch transparently."
"All built on publicly available federal data — proving that meaningful airline intelligence doesn't require proprietary datasets."
- Why not Prophet/LSTM? Only 3 quarters of data. Need 2+ years minimum. Honest about limitations.
- Scope mismatch? P52 covers domestic+international expenses; T-100 is domestic-only. Widebody CASM unreliable. Detected with dual-filter approach.
- 737 vs A320? Prior analyses had regional jets miscoded as A320 family. After correction: virtually identical.
- Business impact? UA could save ~$12M/year on DEN-ORD alone with B737-800→A321 switch.