ONE DATA currently has two types of Models: Machine Learning Models and Process Models.
All available Models in a project can be found under the Models tab. From this view, Models can be uploaded, deleted, shared and moved from and to other projects. To create a model, various processors can be executed in workflows. The Model hub is another way to interact with models.
When creating or uploading Models, note that they need to have a unique name within a Domain in ONE DATA.
Machine Learning Models
The majority of Model use cases in ONE DATA contain Machine Learning Models. These Models can be described as functions f(x, β), with x being input data and β being configurable parameters. The result of these functions is a forecast or classification of the input data. A Model first has to be trained to be able to do a forecast.
Inside ONE DATA, these Models have the Model category "Machine Learning" and the type "MLeap / Spark".
The following processors use Models.
|Processor||Create Model / Add new version||Load and use existing Model|
Under the project VBPM (Virtual Business Process Management) the visualization of business process models was added to ONE DATA. These models are not generated by ONE DATA. They are uploaded as Nimbus .bpm/.xml files and can only be visualized, not adapted. They are listed with the Model category "PROCESS_MODEL" and type "BPM_XCHANGE".
Models can be visualized using Reports. "Add Report Container" in the side bar of the Report view opens the dialog seen in the following image. The option "Model" creates a container, for which the Model to be visualized needs to be selected. A suitable chart type (e.g. Turtle Chart) can then be chosen for visualization.
The user does not have much influence with Reports. Data enrichment is not possible and visualization is limited. More options and possibilities are available when using Apps. It is therefore recommended to use Apps for process model visualization.
- Model Groups (for more advanced Users)