Data Science

With the abundance of data available historically and real-time there is an opportunity to better use the data for the day to day decision making processes in the asset. IPCOS distinguishes four pillars where data science can be applied and where different types of analytics are applicable.

Traditional Data Science approaches face multiple challenges

  • Modeling approaches are too complex to build or maintain
  • The relationship between rigorous models and data driven models is unclear
  • Data scientists are specialists in the latest scientific techniques but have little understanding of the domain

How did we perform and why?

  • Capture and store data
  • Perform root cause analysis
  • Derive statistics and analytical models

How can we maximize performance now?
Where is my attention needed?

  • Automated anomaly detection
  • Orchestration of business processes
  • Real-time optimization

How will I perform in the near future? How can I improve?

  • Prediction of risk / failures
  • Integrated planning
  • Condition-based maintenance

How will I perform in the long-term future?

  • Assets benchmarking
  • Opportunity evaluation
  • What-if scenario’s

Best practices

  • Use open source if available and appropriate
  • Use fit for purpose models with the right level of degree of freedom with respect to the available data
  • Use heuristics when necessary to improve usability and applicability.
  • Don’t sell prove of concepts as products