Create a Butterfly mining run

The Butterfly Mining run type shows the flow in and out of a screen. Use this type to identify all the paths that include this screen, as well as the steps before and after that screen.

Procedure

  1. Navigate to Data > Mining Runs.
  2. Click New, navigate to the Advanced tab, and provide the following information:
    Field Description
    Mining level Ensure Screen Signature is selected
    Mining type Ensure Butterfly is selected
    Filters Create rules to filter which of the recorded screens to include in this mining run.
    Center Screen Signature: Screen signature Specify a screen that all the paths must include.

    The screen signature is a descriptive name assigned to each recorded screen. You can view and search for screen signatures from the Data > Events tab.

    Center Screen Signature: Signature label

    Field: The name of the field the user interacted with

    Control key: The keyboard or mouse control key the user pressed, such as a left click or "Enter"

    You can find field and control key data in the Data > Events page.

    Must Include Screen Signature (optional) Include additional screen signatures that the paths must contain
    Number of events before center The number of steps before the center screen to include in the paths
    Number of events after center The number of steps after the center screen to include in the paths
    Number of processes Sets a maximum number of unique starting screens
    Number of paths Sets a maximum number of total paths
    Start strategy If you do not provide a start screen, Process Discovery uses one of the following algorithms to determine potential start screens:
    • Dependency: Uses a statistical model based on the frequency that certain events lead to one-another or to many others
    • Frequency: Returns the screen signatures appearing most frequently in the event logs for the given filter rules
    • Interactivity: Uses a ratio of unique appearances of screens to the amount those screens are interacted with to determine good start candidates
    • Copy Paste: Looks at controls to determine which screens are copied from most frequently
    Path filter
    • Frequency: Ranks paths by total frequency of occurrence
    • Levenshtein: Ranks paths based on frequency and similarity to other frequently appearing paths
    • Dedup: Includes only the first occurrence of a screen in a unique path, then runs the default filtering on the results. For example, if a process flows through these screens [A -> B -> A -> C], it is converted to [A -> B-> C]
  3. Click Create & Run.
    Process Discovery performs the mining run. The length of time necessary to complete the mining run depends on the size of the data set. The output is a sankey chart, which provides a visual representation of the paths through the selected signature screen.An example of the resulting sankey chat

Next steps

Review the processes identified by the mining run. See Analyze and prioritize processes.