Integrating multiple databases to improve train composition

Integrate multiple structured databases into one coherent model
Logo Ypto: Belgian Railway Company

Pain Points

  • Multiple databases
  • Duplicated and missing data
  • Difficult data integration
  • Missing information affecting train composition


  • Data not fully integrated
  • Data widely dispersed
  • Master Data Management
  • Relevant links hard to find


  • One integrated model
  • Self extending
  • Can be queried without knowing the model itself


  • Fully deduplicated data
  • No data conflicts
  • Near-real time insights at a macro and micro level
  • Data model about 50% smaller

Operation Results

  • Improved insight into train composition
  • New options for seat reservations
  • Better management of capacity and delays
  • Optimization of non-commercial journeys
  • Easier data management and analytics


  • 12 weeks total
  • 4 weeks data gathering
  • 2 weeks basic modelling
  • 2 weeks building queries
  • 3 weeks extending and testing the model
  • 1 week building dashboards