Productionization

During model development, Data Scientists need to test their models with unseen data from database connections within WebFocus. This is a repeating process as changes in during development are frequent. Data Scientists are ones who know best their models and algorithms, code version, data version, and model version, so should have write access to the corresponding app folder in WebFocus dev environment to do this task efficiently.

During development at the time of testing the model in WebFocus dev environment, Data Scientists need to

  • Pull the selected model version with best performance from dvc storage
  • Create/Pull the sample dataset from dvc storage
  • Create/Update the script that loads the model from Bitbucket, then copy it to Dev environment
    • Commit and push the code afterward
  • Create/Update master files
    • Commit and push the code afterward
  • Create/Update fex files
    • Commit and push the code afterward
  • Create/Update reports/charts/dashboards

In UAT, Data Scientists copy files to the UAT environment, and have their customers checked them out.

Productionization via WebFocus may include steps:

  • WebFocus System Administrators get the files in the UAT environment and copy them to the Production environment