Eurail Data Expo 2024

Transcript

1. Andrew Martinez – Novoa Head of Business Intelligence Naveen Nagavelly​ ​ ​ ​ Senior BI Engineer Katerina Tashoska Cloud Architect Data: The Fuel Powering Train Travel
2. Eurail: Helping travelers to discover Europe • Founded in 1959 and based in Utrecht. • Collaborate with over 35 European railway and ferry companies. • Eurail offers rail passes allowing travelers to hop on and off trains and explore 33 European countries. • We provide flexibility and freedom to explore at your own pace. September 13, 2024 2
3. 13 September 2024 3 Eurail: Revolutionizing Railways • We want to: • Use Customer Psychological Profiling to understand our customers and provide them with new products and services they don’t even know they want or need yet • Predict top city destinations and routes each year and work with carriers to make sure there are enough trains and seats for our travelers • Create more complex sales forecasting using machine learning to predict revenue so we can plan initiatives and costs ahead of time even better
4. 13 September 2024 4 Background into Eurail’s Database Systems • Eurail has an on-premises SQL server database where we are querying and ingesting data. • We also have a variety of data including Sales, Reservations, Payment, Passes, Travel, NPS etc… • Different levels of data normalization and data structuring Why move to the cloud?
5. 13 September 2024 5 Eurail: the one-stop shop A Good Database has: • Data Integrity • Data Security • Scalability • Performance • Reliability • Accessibility Our Challenges: • Data Integrity • Scalability • Performance
6. 13 September 2024 6 Data: Unlocking Insights and Driving Decisions • Started with a POC (proof of concept) with small data set. • Integration and migration with end-to-end testing • We also considered maintenance and estimated cost. • Looked all edge cases and decided to move to cloud.
7. 13 September 2024 7 Data: How Data Transforms to Train Operations What are the expectations of having the data in the cloud? • Scalability – Seamless storing of data and processing capabilities • Event-driven – Push bucketing • Accessibility - Can be accessed any time and anywhere • Security - Compliant and with strong data governance • Reliability - High availability with no or minimum downtime • Data Management – Clean and accurate data • Performance - Fast • Opportunity for the implementation of machine learning
8. 13 September 2024 8 What we learned from this project? • Assessment and Planning • Migration Strategies • Data Security and Compliance • Testing and Validation • Optimization and Tuning • Monitoring • Training and Documentation • Post-Migration Review
9. Solution: Modernization
10. Solution: BI platform
11. Solution: Why a Data Lake? • Structure freedom: store unstructed, semi-structured and/or structured data • Analytical • Schema on read • Agile: freedom to define use cases • Low-cost 13 September 2024 11
12. 13 September 2024 12 Q & A