Success Story

Rail ETA Prediction Solution for a Service Provider for the North American Freight Railway Industry

Business Problem

  • ETA accuracy is very important to rail customers as it has significant cost implications
  • ETA very difficult to estimate as per traditional methods due to complexity and noise


  • Online machine learning
  • Transit models
  • Dwell models
  • Routing and Estimation
  • Graph based routing

Business Outcomes

More accurate ETA prediction

Scope of Engagement

3 Member multi-disciplinary team / 6 months

Services Provided

  • Data Preparation for analytics and model development
  • Technologies used: Spark MLLib, Spark streaming, MemSQL and neo4j