Model Simulation View

This component generates a view to interactively execute a model on an artificial data point. The view updates a visualization of the model output based on a new input manually inserted by the user. Adopt the view to either test a hypothesis or to deploy the model as a data app on KNIME WebPortal.

The component works for any classification or regression model captured in a workflow object via the KNIME Integrated Deployment Extension. Supported types for the input feature columns are: “Number (double)”, “Number (integer)” and “String”. For testing this component we recommend connecting its input model port with the “AutoML” classification component (kni.me/c/33fQGaQzuZByy6hE) or the “AutoML (Regression)” component (https://kni.me/c/5kzQcySUa8oukv0Y).

In order to work you also need to provide a sample of data with all the input feature columns of the model. The sample is used to compute the range of possible values (upper/lower bounds and string categories) that the user can change in the view. Constant numeric columns are not supported.

For deployment purposes you can also save the last input and prediction by adding more nodes to the component output.

This component was released as part of the Verified Components project (knime.com/verified-components).

Options

Select Feature Columns

Select all and only the columns that are the input of your model. Please make sure any additional column(s) which would be ignored by the model is excluded. For example exclude the 'target' or 'ground truth' usually attached to the train and test set. Only 'Number (integer)', 'Number (double)' and 'String' types are supported.

Select Model Type

This view supports two types of ML model: classification models (both binary and multiclass) and regression models. Please select which type of model you connected at the input. If your model predicts strings/categorical target you should select "classification". If your model predicts a numeric target you should select "regression".

Cache Files ID

The identifier used to save in the workflow data area KNIME tables as cache for consistent interactions. If multiple instances of this component are executed in parallel, make sure to set up different identifiers.

Nodes

• Empty Table Switch65 ×
• Transpose60 ×
• Table Row to Variable59 ×
• Column Rename58 ×
• Component Input56 ×
• Show all 59 nodes

Extensions

• FeatureKNIME Base nodes
• FeatureKNIME Integrated Deployment
• FeatureKNIME JavaScript Views
• FeatureKNIME Javasnippet
• FeatureKNIME Math Expression (JEP)
• Show all 6 modules

Links

• No links available