Kevin O'Sullivan, SITA Lab, presents at SITA 2013 Europe Aviation ICT Forum
1 like1,595 views
This document summarizes a project conducted by SITA Lab at Sydney Airport to analyze big data and predict passenger flows. It describes how they used WiFi analytics, flight schedules, FIDS data, and immigration data to predict arrivals flows and provide recommendations. Key learnings included focusing on business intelligence objectives rather than just analyzing data, using commodity cloud servers and open source software to start small, and prioritizing learning over specialized hardware or experts. The goal of telling stakeholders something new about passenger flows that they did not already know from this big data analysis was achieved.
Kevin O'Sullivan, SITA Lab, presents at SITA 2013 Europe Aviation ICT Forum
- 1. SITA LAB
- 2. Sydney Airport Big Data Tell me something I don’t know. 17th October SITA Lab SITA LAB
- 3. Contents • CPH History / Big Data Gartner • Sydney Project Overview • What did we learn • Conclusion SITA LAB 3 | Sydney Analytics | Confidential | © SITA 2012
- 4. Copenhagen Airport Pilot (2011) World premiere of passenger flow management via WIFI SITA LAB 4 | Innovating Together|iFlow Confidential | © SITA 2012
- 5. Google Trends, search term “Big Data” SITA LAB 5 | Sydney Analytics | Confidential | © SITA 2012
- 6. Hype Cycle for Emerging Technologies 2013 Source: Gartner SITA LAB 6 | Sydney Analytics | Confidential | © SITA 2012
- 7. SITA LAB 7 | Sydney Analytics | Confidential | © SITA 2012
- 8. Sydney Airport Big Data Objectives • Predict accurate passenger flows across airport key facilitation zones, and in particular during an anomaly. • E.g. multiple large body aircraft unexpectedly arriving at same time. i.e. Tell me something I don’t know SITA LAB 8 | Sydney Analytics | Confidential | © SITA 2012
- 9. Outcomes for Arrivals 2 1 Predicted Flows 3 4 Roster Recommendations 6 Exit Flow Forecast SITA LAB 9 Total Pax Counts | data science engine | Confidential SITA | © SITA Transit Gate Estimations
- 10. Data Sources Used WiFi Analytics from iFlow Used to measure time to walk from gates to immigration and verify predictions Flight Schedules Used to get scheduled flight data. FIDS data Used get estimated arrival times and passenger counts Immigration Data Used to measure processing time at immigration SITA LAB 10 | Sydney Analytics | Confidential SITA | © SITA
- 11. What does Big Data architecture look like? Data Sources Visualization & Alerting HTML5 ETL email/SMS Prediction Engine R Java Hadoop Data Grid Oracle Big Data Appliance Cassandra SITA LAB 11 | Sydney Analytics | Confidential SITA | © SITA HPCC IBM Big Insights
- 12. Google Glass Alert SITA LAB
- 13. What did we learn? • Big Data projects are not about the data. • Big Data projects are about Business Intelligence. • Don’t be lured into conducting a fishing exercise on your data. • Set the BI objects at start of project, and stay lazer focused on these objectives. SITA LAB 13 | Sydney Analytics | Confidential SITA | © SITA
- 14. What did we learn? • You don’t need to commit big dollars for dedicated hardware and specalist software. • Run a starter project on commodity cloud servers and open source software. • Learn by doing. • It maybe cheaper for existing staff to learn big data, than a big data expert to learn your business SITA LAB 14 | Sydney Analytics | Confidential SITA | © SITA
- 15. Conclusion • Start small with Big Data • Focus on the Business Intelligence • Tell me something I don’t know. SITA LAB 15 | Sydney Analytics | Confidential SITA | © SITA