Model Management

In modern digital architectures, descriptive and data driven models play an important role. They are used heavily in a number of workflows such as virtual metering, data reconciliation, optimization and anomaly detection. Managing and maintaining the models in a structured way is imperative to scale and sustain the digital solution in the long run. IPCOS has built up extensive experience in designing and implementing model management solutions both using commercial off the shelve products as well as custom developed tools.

IPCOS has developed a reference architecture which clarifies the role of the model management solution:

IPCOS’ Simplified reference architecture
IPCOS’ Simplified reference architecture

Model Management overcomes multiple challenges

  • There is no single version of the model – models are kept privately by the engineer or are stored on a shared drive
  • There is no version control on the models and it is impossible to trace back who modified the model for what purpose
  • Engineers lose lots of time manually updating the model resulting lots of unmaintained models
  • There is no clear workflow on how to update the models and what procedure to follow

Best practices

  • Model Management solutions ensure proper storage and versioning of the various models in use
  • Models are stored in model repositories ensuring full availability for all workflows in the digital architecture
  • Model repositories follow a structure consistent with the data model to enable easy mapping between the models and the data