Leveraging Predictive Analytics to Turn Big Data into Operations Improvement

Transcription

1 Leveraging Predictive Analytics to Turn Big Data into Operations Improvement Pete Nelson Solutions Architect, Quintiq Radnor, PA

2 Key presentation takeaways Key presentation takeaways Predictive analytics in operations Using predictive analytics to improve scheduling Best practices in implementing predictive analytics

3 About Quintiq Develop a single application capable of solving any type of planning puzzle > 1,000 employees Founded in 1997 Used by companies in over 80 countries > 300 customers worldwide 12,000 users A Dassault Systèmes company since 2014

4 Quintiq in Rail Long Short Day of operation Long term Standard week Anonymous plan Short term Actual week Assigned to resources Day of ops Disruption management Best use of available resources Timetable Quintiq in Rail Timetable Short term timetable Delay and disruptions Consist Unit diagramming Unit planning Unit swapping Crew Crew diagramming Crew roster Crew swapping Maintenance Resource diagramming Resource schedule Resource swapping

5 The explosion in public transport data The explosion of data in public transport Trend Effect on Operations Internet of Things (IoT) Track conditions The changing landscape of rail Rolling stock conditions operations Operational and asset systems Fare and revenue systems Time and attendance Maintenance and task durations Service performance Ridership data Revenue trends

6 The stages of advanced analytics The evolution of advanced analytics What happened? Why did it happen? What will happen? How can we make it happen? Descriptive Analytics Diagnostic Analytics Predictive Analytics Prescriptive Analytics INFORMATION OPTIMIZATION

7 Gaining value from past performance Gaining value from past performance D e v e l o p b a s e l i n e k n o w l e d g e Execution Execute the current plan Feedback loop Capture actuals Capture the actuals Specify the variable to predict Update knowledge Feed recent realization data into the machine learning engine Generate an up-to-date predictive model

8 Capturing actual dwell times

9 Predictive analysis of dwell times

10 Predictive analytics in maintenance planning Use case: Maintenance planning Average time per task Actual Predicted Planned

11 Where Where is is the the value value in in rail rail operations? operations? Timetables Consist and Crew Planning Maintenance Inputs Trip durations between stations Dwell times Ridership data Time and attendance Maintenance durations Value More efficient service scheduling Improved allocation of resources Visibility into hours of service violations More accurate maintenance window planning

12 Keys to implementing predictive analytics Keys to implementing Predictive Analytics Determine where the value lies Don t overlook the human touch of planners Relating cause and effect don t learn bad practices

13 Closing the loop Closing the loop Timetables Time & attendance Consist planning Powered by real-world data Day of operations Crew planning Self-learning planning and scheduling system Mobility Leave & training planner Maintenance planning

14 Thank you! Join us tonight at the Quintiq Reception Latrobe Room, 6-8PM Pete Nelson Solutions Architect, Quintiq