Skip to content

Model management

Model entity

In the TranzAI platform, a model is a very simple entity that specifies:

  • the target variable that should be calculated or predicted
  • the type of model (regression, classification ...)
  • the training data set that is used to train model versions
  • the set of features available for model design

From this entity, data scientists can launch feature selection processes and create different model versions that will be trained and evaluated to select the version that best match each use case requirements:

  • explainability and transparency
  • accuracy
  • stability
  • operationalization (modelOps)

Model documentation

In TranzAI, all tasks associated with the iterative process of designing, training, and testing a model are automatically recorded and self-documented thanks to the metadata-driven nature of the TranzAI platform.

Training environments

Training notebooks run indepently with an environment that is specified at runtime.

The TranzAI platform automatically provision the environment required to train a model according to the type of algorithm and libraries you used when designing your model.