Selecting Parts of a Process with Select Flow

When you are working with complex processes you may want to use different elements from different parts or branches of the process together.

Rulex provides a task, called Select Flows from which this operation can be performed.

For example, you may have two separate processes: one used to train the model, and a second process used for scoring. The first process may be executed on a monthly basis, while the second scoring process on a daily basis. You can import the training process into the scoring process and use the Select Flows task to merge the two flows exactly as you require, taking elements, such as datasets, clusters and rules, from the required task.

Prerequisites

Additional tabs

The following additional tabs are included in the task:


Procedure

  1. Drag and drop the Select Flows task onto the stage.

  2. Connect the tasks whose elements you want to use to the new task.

  3. Double click the Select Flows task.

  4. Configure the options described in the Select Flows options table below.

  5. Save and compute the task.

Select Flow options

Parameter Name

PO

Description

Get dataset from

datatask

Select the task from the drop-down from which the dataset will be taken.

Get rules and relevances from

ruletask

Select the task from the drop-down from which rules and relevances will be taken.

Get black-box model from

modtask

Select the task from the drop-down from which black-box models will be taken.

Get clusters from

clusttask

Select the task from the drop-down from which clusters will be taken.

Get association rules from

asrultask

Select the task from the drop-down from which association rules will be taken.

Get discretization cutoffs from

cutofftask

Select the task from the drop-down from which discretization cutoffs will be taken.

Get pca eigenvectors from

pcatask

Select the task from the drop-down from which PCA eigenvectors will be taken.

Get auto regressive models from

autoregtask

Select the task from the drop-down from which auto regressive models will be taken.

Recompute covering error (if possible)

forcecover

If selected, covering and error attributes are recomputed, if a dataset is present.