Symptoms and root causes
Symptoms are observed when a process analysis is performed, whereby the processes are analyzed with the help of a base set of filters.
For example, you inspect your global sales processes of the last year and you find that a disproportionately large number of customer complaints occur in your subsidiary in China compared to the rest of the world. You start a process analysis to find out the cause of this asymmetry. In addition to the filters that define the base set of your observation (here: sales processes, last year’s processes, and country) you specify the dimension value that you want an explanation for (here: complaints). This dimension value is called the symptom. PPM uses your settings to analyze all suitable dimensions for values with strikingly high or low numbers of occurrences in combination with the symptom. The result of the analysis is a table that lists all dimension values together with an estimate of how strongly this value affects the occurrence of the symptomatic value. This table may contain a row showing that complaints very often occur together with the value "PN4711" for the product dimension. The dimension value "PN4711" is called a root cause.
The following criteria can be used as symptoms.
Single-level dimension value, as mentioned in the example
Multi-level dimensions, which are interpreted disjunctively, such as the dimension Equipment with the values basic or luxury.
KPIs
Value ranges, for example, Processing time with the range 10 to 12 hours. When the symptom contains a process measure, all process instances for which that measure is undefined are counted among the non-symptomatic processes.
The following criteria are not supported.
Time range dimensions
Function and relation dimensions. The symptom must be specified on process level.
Supported dimensions
Only single-level text dimensions on process level and the variant dimension are analyzed as root causes.
Variants are analyzed on combined and precise level independently.
Only dimensions that are available in the Process Mining context are included in the analysis.
Non-supported dimensions
Function and relation dimensions are not analyzed.
Any dimension that already occurs in the symptom is ignored.
Dimensions that contain user-defined steps or usages are ignored.