SlideShare a Scribd company logo

Schiphol - Optimale doorstroom van passagiers op Schiphol dankzij slimme datatoepassingen

•
•
B
BigDataExpo

Schiphol is Europa’s best connected airport en verwerkt op piekdagen tot 235.000 passagiers. Om deze soepel door de processen te leiden is een betrouwbare prognose van de drukte noodzakelijk. Schiphol laat zien hoe zij datatoepassingen ontwikkelt om het aantal reizigers zo accuraat mogelijk te voorspellen en hiermee processen in te richten.

Related slideshows

Flight data analysis using apache pig--------------Final Year Project
Flight data analysis using apache pig--------------Final Year ProjectFlight data analysis using apache pig--------------Final Year Project
Flight data analysis using apache pig--------------Final Year Project
 
The Perfect Travel and Expense Management Checklist
The Perfect Travel and Expense Management ChecklistThe Perfect Travel and Expense Management Checklist
The Perfect Travel and Expense Management Checklist
 
Reading in the future2
Reading in the future2Reading in the future2
Reading in the future2
 
1 of 23
Download to read offline
Optimale doorstroom van passagiers op Schiphol dankzij slimme datatoepassingen 20 SEP 2018 Marcel Raas Berend Onnes &
Agenda 2 Introduction • About Schiphol • Traffic Analysis & Forecasting team Implementation of machine learning • Operational forecasting • Tough ride to get it implemented • What did/do we do to make it a success? Questions
Some big numbers (before talking big data) 3
Traffic Analysis & Forecasting Team 4 Reporting & Analysis • Network analysis (connectivity) • Benchmarking • Opportunities / threats • Regional Airports Capacity • Forecasting Short term => hiring staff Mid/Long term => infra planning Long term => spatial planning • Limitations: airport slots, airport & airspace capacity, environment, etc. Commercial • Airline business cases: Route / business development • Cargo cases • Forecasts to set airport charges € €
Agenda 5 Introduction • About Schiphol • Traffic Analysis & Forecasting team Implementation of machine learning • Operational forecasting • Tough ride to get it implemented • What did/do we do to make it a success? Questions
Confidential 6 Importance of forecasting
Confidential 7 Forecasting Flight Schedules Time Destination Aircraft type # Seats Airline Slot-file Market insights ProductDataSource Passengers per movement Added: # passengers Transfer share Airline Planning tool Market insights Flow forecast Added: Routing Reporting patterns Ticket scans Operational insights Planning tool Staff planning Added: Throughput = ‘productivity’ Historic throughput LOWWAITINGTIMES @LOWCOSTS 1 2 3 4
Confidential 8 Lazy / Lean… Why would an Airport work so hard to predict, while Airlines… • Have insights in sales & searches • ‘manage their yields’ (start promotion / discounts) have best insights on the expected number of passengers per flight Increase the share of airline input
Confidential 9 … but stay in control… Avoid full dependency of airlines Validate Airlines’ input (no strategic behaviour?) Strong tooling
Confidencelevel Time 10 Not user friendly and a Lack of faith in ‘machine’ Results were strong, now get organized professionally Make the best & full blown Looks great! Now make it robust Building a GUI is not our strength The ML engine is OK, though btw, legacy tool isn’t perfect either That last 5% is tough Promising results Forget about bureaucracy & go! … hence we developed a tool using machine learning
Confidential 11 How to build a robust model?
Confidential Predict Train • Predict the past • Make it realistic: do not use future information Feedback: measure how well it performed How to build a robust model?
Confidential • What machine learning algorithm to use • What data to include and what not • How to model holdiays • Importance weighting of more recent data/holidays What did we tune? 13
Confidential Dealing with the Dutch holidays 14 • Holidays appear in a lot of different configurations • Employ time-series analysis and disentangle holidays per region • These predictions: additional inputs to machine learning algorithm • Final tweak: weights, more importance to holiday periods Central +20% North + 15% South + 5%
Adding weights • Holiday periods • Aircraft size • Historic years • Winter/Summer season trends Weights for Summer 2017 training
Measure whether it improved 16 Sum of all flights departing in the same 15 minute bracket ML prediction too highML prediction too low Test data
When is it ‘good enough to go’? 17 Daily totals Grouped by airline Grouped by destination Single flight level
Confidential Reliable training data doesn’t always appear accurate Don’t rely on historic data to be fully accurate. Things we revealed in our historic data • 100 seats and…. 160 passengers • -/- 61 passengers to Mumbai • Freight shipped from Amsterdam to… Amsterdam!
Measure forecast accuracy Total Local Transfer Test data
Be thankful ☺☺☺☺ for cooperation Test data
See where you can improve Schedule Load factor Transfer share # seats Test data
ML Roller coaster learnings 22 Got organized seriously Define “when is it good enough to go?” Hired skilled professionals New tools = new user skills Compare & validate Data accuracy
Agenda 23 Introduction • About Schiphol • Traffic Analysis & Forecasting team Implementation of machine learning • Operational forecasting • Tough ride to get it implemented • What did/do we do to make it a success? Questions

More Related Content

Schiphol - Optimale doorstroom van passagiers op Schiphol dankzij slimme datatoepassingen

  • 1. Optimale doorstroom van passagiers op Schiphol dankzij slimme datatoepassingen 20 SEP 2018 Marcel Raas Berend Onnes &
  • 2. Agenda 2 Introduction • About Schiphol • Traffic Analysis & Forecasting team Implementation of machine learning • Operational forecasting • Tough ride to get it implemented • What did/do we do to make it a success? Questions
  • 3. Some big numbers (before talking big data) 3
  • 4. Traffic Analysis & Forecasting Team 4 Reporting & Analysis • Network analysis (connectivity) • Benchmarking • Opportunities / threats • Regional Airports Capacity • Forecasting Short term => hiring staff Mid/Long term => infra planning Long term => spatial planning • Limitations: airport slots, airport & airspace capacity, environment, etc. Commercial • Airline business cases: Route / business development • Cargo cases • Forecasts to set airport charges € €
  • 5. Agenda 5 Introduction • About Schiphol • Traffic Analysis & Forecasting team Implementation of machine learning • Operational forecasting • Tough ride to get it implemented • What did/do we do to make it a success? Questions
  • 7. Confidential 7 Forecasting Flight Schedules Time Destination Aircraft type # Seats Airline Slot-file Market insights ProductDataSource Passengers per movement Added: # passengers Transfer share Airline Planning tool Market insights Flow forecast Added: Routing Reporting patterns Ticket scans Operational insights Planning tool Staff planning Added: Throughput = ‘productivity’ Historic throughput LOWWAITINGTIMES @LOWCOSTS 1 2 3 4
  • 8. Confidential 8 Lazy / Lean… Why would an Airport work so hard to predict, while Airlines… • Have insights in sales & searches • ‘manage their yields’ (start promotion / discounts) have best insights on the expected number of passengers per flight Increase the share of airline input
  • 9. Confidential 9 … but stay in control… Avoid full dependency of airlines Validate Airlines’ input (no strategic behaviour?) Strong tooling
  • 10. Confidencelevel Time 10 Not user friendly and a Lack of faith in ‘machine’ Results were strong, now get organized professionally Make the best & full blown Looks great! Now make it robust Building a GUI is not our strength The ML engine is OK, though btw, legacy tool isn’t perfect either That last 5% is tough Promising results Forget about bureaucracy & go! … hence we developed a tool using machine learning
  • 11. Confidential 11 How to build a robust model?
  • 12. Confidential Predict Train • Predict the past • Make it realistic: do not use future information Feedback: measure how well it performed How to build a robust model?
  • 13. Confidential • What machine learning algorithm to use • What data to include and what not • How to model holdiays • Importance weighting of more recent data/holidays What did we tune? 13
  • 14. Confidential Dealing with the Dutch holidays 14 • Holidays appear in a lot of different configurations • Employ time-series analysis and disentangle holidays per region • These predictions: additional inputs to machine learning algorithm • Final tweak: weights, more importance to holiday periods Central +20% North + 15% South + 5%
  • 15. Adding weights • Holiday periods • Aircraft size • Historic years • Winter/Summer season trends Weights for Summer 2017 training
  • 16. Measure whether it improved 16 Sum of all flights departing in the same 15 minute bracket ML prediction too highML prediction too low Test data
  • 17. When is it ‘good enough to go’? 17 Daily totals Grouped by airline Grouped by destination Single flight level
  • 18. Confidential Reliable training data doesn’t always appear accurate Don’t rely on historic data to be fully accurate. Things we revealed in our historic data • 100 seats and…. 160 passengers • -/- 61 passengers to Mumbai • Freight shipped from Amsterdam to… Amsterdam!
  • 20. Be thankful ☺☺☺☺ for cooperation Test data
  • 21. See where you can improve Schedule Load factor Transfer share # seats Test data
  • 22. ML Roller coaster learnings 22 Got organized seriously Define “when is it good enough to go?” Hired skilled professionals New tools = new user skills Compare & validate Data accuracy
  • 23. Agenda 23 Introduction • About Schiphol • Traffic Analysis & Forecasting team Implementation of machine learning • Operational forecasting • Tough ride to get it implemented • What did/do we do to make it a success? Questions