© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data Driven Public
Transport
Saku Vaittinen, VR Group | 21.3.2018
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
VR Group: Moving people and other things
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
VR Group’s track to AWS
• Cost effectiveness + elasticity + development speed
• First implementations 2016, core infra ready 2017 (CON206)
• Terraform
• Direct Connect
• Shift from on-prem to cloud required lots from organization
• Comprehensive mindset change with DevOps through new projects
• Lift’n’shift avoided
• Focus on architecture!
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data on driver’s seat
• Incentive: Free competition on local railway passenger traffic by
2021 and on long-haul by 2024
• From Excel to Data science
• Learning from data
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Analytics project
• Started in January 2018
• 6 members, everyone new
• Lots of legacy burden
• Almost production-ready
• Infrastructure, Data platform, Data model
• First data sources
• Power BI dashboards and reports
• Recommendation API for new ticket system
• Data science work has been started
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
OPERATIVE INTELLIGENCE CUSTOMER INTELLIGENCE
RESOURCE
PLANNING
PERSONALIZED CONTENT
PRODUCT, PRICING
STRATEGY
CUSTOMER SERVICE
ROBOTIZATION
CORPORATE
TRAVEL
CUSTOMER
VALUE & CHURN
SHIFT PLANNING
OTHER OPERATION
OPTIMIZATION
LOYALTY
PROGRAM
DEVELOPMENT
The Problem
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Diverse data ecosystem
Data sources
• DBs: on-premise, cloud
• Apps: mobile, web, ticket machines
• Systems: seat reservation, train planning, ERP
• Communications: email, surveys, social media
Data users
• Dashboards & analytics applications
• Personalized content
• Operative intelligence
• Customer service
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Data analytics stack
Cloud
• AWS
Open source tools
• Apache NiFi, Jupyter, DbSchema, Python …
Business analytics applications
• Power BI
Warehouse
• Snowflake
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
AWS Architecture
Connectivity
• VPC Privatelink
• Direct in-account connection to Snowflake in same region
Data pipeline
• Data Migration Service
• Daily batches, Continuous migration by CDC files
Analytics API
• ECS, Lambda, API GW, ElastiCache
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Snowflake characteristics
• AWS-native
• Pay-per-use pricing model
• Separated storage & compute
• Virtual Data Warehouses
• Variable patterns both in data import & usage
• Batch, stream
• BI users, Data science, API
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
© 2018, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
Next hot things
• Raw data from ALL sources
• Machine Learning for e.g.:
• Time series forecast algorithms
• Ticket pricing
• Work shift lists
• Train configuration planning
• Immediate reaction to changes

AWS Community Day Nordics 2018 - Saku Vaittinen (VR): Data driven public transport

  • 1.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data Driven Public Transport Saku Vaittinen, VR Group | 21.3.2018
  • 2.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. VR Group: Moving people and other things
  • 3.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. VR Group’s track to AWS • Cost effectiveness + elasticity + development speed • First implementations 2016, core infra ready 2017 (CON206) • Terraform • Direct Connect • Shift from on-prem to cloud required lots from organization • Comprehensive mindset change with DevOps through new projects • Lift’n’shift avoided • Focus on architecture!
  • 4.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data on driver’s seat • Incentive: Free competition on local railway passenger traffic by 2021 and on long-haul by 2024 • From Excel to Data science • Learning from data
  • 5.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Analytics project • Started in January 2018 • 6 members, everyone new • Lots of legacy burden • Almost production-ready • Infrastructure, Data platform, Data model • First data sources • Power BI dashboards and reports • Recommendation API for new ticket system • Data science work has been started
  • 6.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. OPERATIVE INTELLIGENCE CUSTOMER INTELLIGENCE RESOURCE PLANNING PERSONALIZED CONTENT PRODUCT, PRICING STRATEGY CUSTOMER SERVICE ROBOTIZATION CORPORATE TRAVEL CUSTOMER VALUE & CHURN SHIFT PLANNING OTHER OPERATION OPTIMIZATION LOYALTY PROGRAM DEVELOPMENT The Problem
  • 7.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Diverse data ecosystem Data sources • DBs: on-premise, cloud • Apps: mobile, web, ticket machines • Systems: seat reservation, train planning, ERP • Communications: email, surveys, social media Data users • Dashboards & analytics applications • Personalized content • Operative intelligence • Customer service
  • 8.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Data analytics stack Cloud • AWS Open source tools • Apache NiFi, Jupyter, DbSchema, Python … Business analytics applications • Power BI Warehouse • Snowflake
  • 9.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. AWS Architecture Connectivity • VPC Privatelink • Direct in-account connection to Snowflake in same region Data pipeline • Data Migration Service • Daily batches, Continuous migration by CDC files Analytics API • ECS, Lambda, API GW, ElastiCache
  • 10.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Snowflake characteristics • AWS-native • Pay-per-use pricing model • Separated storage & compute • Virtual Data Warehouses • Variable patterns both in data import & usage • Batch, stream • BI users, Data science, API
  • 11.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved.
  • 12.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved.
  • 13.
    © 2018, AmazonWeb Services, Inc. or its Affiliates. All rights reserved. Next hot things • Raw data from ALL sources • Machine Learning for e.g.: • Time series forecast algorithms • Ticket pricing • Work shift lists • Train configuration planning • Immediate reaction to changes