Back to all cases Preventing incidents using a data-driven tool at ProRail
ProRail manages the 7,000 kilometer Dutch railway network. To increase capacity and reduce delays, ProRail is transitioning with the help of BigData Republic from a reactive approach to a data driven, preventive deployment of the Incident Response Team (ICB).
The challenge
A significant proportion of railway incidents is caused by third parties (such as track trespassers, vandalism, and loose dogs). The ICB team, comprising approximately 200 FTEs, previously operated primarily in a reactive manner. To enhance safety and increase passenger capacity by 30%, ProRail sought to actively prevent incidents using data.
The solution
A multidisciplinary project team (including machine learning engineers from BigData Republic) developed a data-driven approach:
Data Pipelines & Infrastructure: Setting up automated ETL pipelines using Azure Kubernetes, Docker, and SQL databases.
Power BI Dashboards: Developing insightful dashboards to visually analyze incident patterns and trends.
NLP & Predictive Modeling: Applying Natural Language Processing (NLP) and text mining to historical incident descriptions, combined with predictive machine learning models via AzureML.
Agile Collaboration: Working according to the Scrum methodology, featuring periodic feedback sessions and close stakeholder involvement.
The result
- Weekly planning via dashboards
- High-quality database
- Value from text data
- Embedding & Integration