Active Model
Designate a model in Alviss AI as the default, to automatically be used it in dashboards, simulations, predictions etc.
In Alviss AI, an active model refers to the model that is designated as the default for a specific combination of Country, Region, and Grouping (e.g., "US - West Coast - E-commerce"). This setup ensures that when you run analyses, simulations, predictions, or view dashboards like Effect, the platform automatically uses the active model for that particular combination, streamlining your workflow and reducing the need to manually select models each time.
Activating a model is a simple process that optimizes performance and accessibility:
- It makes the model readily available for generating insights without additional setup.
- Only one model can be active per unique combination at a time; activating a new one will deactivate the previous.
- This feature is especially useful in team environments, where shared projects benefit from consistent defaults across members.
- Lower user roles can only use the default model.
How to Activate a Model
- Navigate to Models: Go to the Models section in your project via the side menu.
- Select a Model: From the list of trained models, choose the one you want to activate.
- Activate: Click the activation option (small button with a lightning bolt). This sets it as the default for its associated combination.
- Optional Automation: During model building in the [Basic Model Builder](Basic Model Builder.md) or [Advanced Model Builder](Advanced Model Builder), you can enable an option to automatically activate the model upon successful training.
Benefits of Using Active Models
- Efficiency: Speeds up insight generation, such as running Attributions, Simulations, or Predictions, by eliminating repetitive selections.
- Consistency: Ensures all team members in a shared project use the same model for a given combination, promoting aligned decision-making.
- Flexibility: You can switch active models as needed, for example, when testing new versions or updating based on fresh data.
Before activating, review the model's performance metrics (e.g., accuracy on hold-out data) to ensure it's the best fit. If you're working with multiple combinations, activate models strategically to cover your key analytical needs.
If your project involves complex setups or you encounter issues with activation, refer to the Models documentation for more details or check your project's settings.