Skip to content

PAFnow Root Cause Analyzer

The PAFnow Root Cause Analyzer helps you to identify the most likely root causes of behaviors, patterns or problems in your process. It is included on Report Pages that relate to specific process characteristics such as the Process Steps Report Page. With the Root Cause Analyzer you can look at the driving forces of positive process behavior or find the root causes of problems and bottlenecks in your process. It is very helpful when making an informed business decision.


The Root Cause Analyzer offers a Compact View and a Scatter View and the five most probable root case compositions can be reviewed individually.

Scatter View


The Scatter View of before seen root cause sorts the bubbles regarding their Precision/Coverage combination.

This is the icon of the Root Cause Analyzer in the Visualizations pane:


How to get started

You need to drag the required fields from the Fields pane into the respective field buckets of the visual in the Visualizations pane (see Figure 2) so the Root Cause Analyser can display the root causes of a process pattern (or process problem).


Count: The field that corresponds to the required level of granularity (Case, Event, Edge Levels) should be dragged into the Count bucket and set to Count in the drop-down-menu (e.g. CaseId from the Cases Table to explain case attributes, or CaseId from the Event Log Table for explaining the event attributes).

Explain By: Relevant attributes (via drag-and-drop) that PAFnow Root Cause Analyzer should consider in order to analyze the root causes. Ideally, you should include attributes that are categorical in nature, such as: Resource, StartActivity, HappyPath, Plant, Vendor, etc. but avoid the ones that are continuous, such as: Duration, Cost, Quantity, etc.

You also might have to adjust your Visual Interactions before the Root Cause Analyzer can work properly.

Now you can start with your first root cause analysis:


In the example, we wanted to identify the root causes for cases with many process steps. We can trigger the Root Cause Analyzer by selecting a bin, for example (g) > 6 process steps from the bar chart on the Variants Report Page.

The PAFnow Root Cause Analyzer displays the top (most likely) root causes for the selection, see Figure 4:

Tooltip Details

When you hover over the bubbles, in both Compact (Figure 5) and Scatter (Figure 6) Views, the tooltip provides detailed information on the root cause.

As shown in the tooltips, 617 cases belong Selection that have more than 6 process steps. And the Rest comprises of 351 cases that have less than 6 process steps. HappyPath:Happy and Has(Self)Loops: 0 explains 473 of the Selection and 103 of the Rest. This root cause has a high Coverage and a hight Precision, leading to a considerably hight score.

With the Ctrl+Click function, you can further drilldown a root cause and analyze its impact on your process in a more detailed visualization.


In the example, we used the Ctrl+Click function on the 79.3 percent bubble. In the drilldown, we can see that 473 cases from the bin (g) > 6 process steps are affected by this root cause (see Figures 7&8).

If you want to analyze the root causes of the Inverse Selection, you can switch to the Inverse Selection Mode on the top right corner of the visual (the red square-button).

Inverse Selection Mode

Following the original example, the root causes of all cases with less than 6 process steps, are shown here.

How to use

Visual Interaction

The Root Cause Analyzer starts its analysis only after you selected a process behavior or problem. For example, selecting a bar from the bar chart, or a portion of the pie from the pie chart, or an edge from the PAFnow Process Viewer would activate the Root Cause Analyzer.


The Root Cause Analyzer works like a selection visual (and not a slicer visual). This means that you have to keep attention to the settings of your visual. The interaction setting of the selection visualization must be set to Highlight, otherwise the Root Cause Analyzer cannot work properly.

To set the interaction to Highlight, choose the visual you want to use as a basis of your root cause analysis. The Formatting pane appears under Visual Tool next to the Help ribbon on top of the canvas. Click Edit Interaction as shown in Figure 10.

Figure 10

This displays the possible settings of the visual’s interaction with all other visuals on your page. The interaction for the Root Cause Analyzer should be set to Highlight (see Figure 11 underneath).


When you hover over the bubbles in Compact or Scatter View (see Figures 5&6), a tooltip with the following information is displayed:

Tooltip Information
Selection The number of data points that have been selected from a selection visual (for example, from bar charts, pie charts, etc.). All the data points in this selection exhibit a similar pattern/behavior.
Rest The number of data points that have not been selected from a selection visual. All the data points in the Rest region exhibit a contradictory pattern/behavior in comparison to the Selection region.
Rule Combination List of attribute(s) and their corresponding value(s) that contribute best as a root cause are listed below the Selection and Rest fields.
Explains Number of data points explained by the rule combination of the Selection and Rest regions.
Coverage Fraction of data points from the Selection that can be accurately explained by the rule combination. In other words, the rule combination should be able to explain the majority of the Selection region in order to qualify as a root cause. This measure is also interchangeably known as Recall.
Precision Fraction of relevant data points (from the Selection region) among the total number of data points explained by the rule combination. For example, if the rule combination explains data points from the Rest region, this would mean that they also contribute to the contradictory pattern observed in the Rest region. This leads to the conclusion that the given rule combination is not precise enough to be called a root cause.

Result Interpretation

Ideally, a strong root cause is a rule combination that has a score close to 100 percent for bubbles displayed in Compact Mode. When switching to Scatter View, a strong root cause positions itself in the top right corner of the grid.

A score displayed on the bubble for a given rule combination is the average of the coverage and precision scores. In addition, the size and color of the bubble is determined by the displayed score.

The stronger a rule combination is, the closer the score is to 100 percent, the bigger the bubble and the darker its color.

The standard setting for the color of a positive root cause is green. However, the color of the root causes - positive, negative and neutral - can be changed in the Formatting pane.


Required Fields

In order to use the Root Cause Analyzer, you have to drag the required fields into the respective buckets.

Name Type Description
Count Numerical Use the CaseId field from the Cases table and choose Count (Distinct) from the drop-down-menu.
Explain by Any Use any attributes from the PAFnow Data Model to explain the selected behavior by.


Following are the various options provided in the Formatting pane formattingIcon that can be used to customize the Root Cause Analyzer.


Name Description
Pause Analysis Any selection action on a report page with an Root Cause Analyzer activates the Root Cause Analyzer. In order to deactivate the visual from responding to every selection, you can press the Pause-button on the visual or slide off the Pause Analysis option from the Formatting Pane as shown in the screenshots.
Multivariate Analysis On enabling this option, multivariate analysis is activated. Multivariate Analysis provides root causes of the selection while considering the interaction effects of multiple attributes.
On the other hand, Univariate Analysis provides root causes of the selection only by considering one attribute value at a time.
View Mode Compact: This view mode displays the root causes in a compact bubble chart.
Scatter: This view mode displays the root causes in a grid view with Coverage and Precision in the horizontal and vertical axes respectively.
View modes can be switched either directly from the visual or through the View Mode settings in the Formatting pane of the visual.
Selection Mode Normal: This mode is used to identify the top five root causes for the subset of data points in the selection.
Invert: This mode is used to identify the top five root causes for the subset of data points outside the selection.
Selection Modes can be switched either directly from the visual or through the View Mode settings in the Formatting pane of the visual.
Neutral Color Color to display neutral/insignificant root causes. Default is grey.
Positive Color Color to display positive root causes. Default is green. Darker green in color denotes a strong positive root cause.
Negative Color Color to display negative root causes (root causes for Inverse Selection). Default is red. Darker red in color denotes a strong negative root cause.
Dynamic Grid This setting changes the grid size based on the size and number of bubbles in the root causes.
Use Transitions This setting, when enabled, is used to animate a smooth transition between Compact and Scatter Views.
Transition Duration Is the field that specifies the time (in milliseconds) it takes to perform the transition.