We can have different kind of metrics :
- A process provides its own metrics mostly based on durations, costs and other information that can be measured directly into the process itself like counts (of repetition, usage of the same event in different variations, etc.) and many other statistics on Process Steps.
- Process Mining Solutions should also enable the creation of other metrics or indicators (KPIs) based on collateral information we would have imported during the previous stage.
- It would also be possible to create some combinations of the metrics used or created above. These combinations can operate
- On the fields (ie. vertically): like a calculation between different attributes of a Process Step.
- On the Process steps themselves (ie. horizontally): like an aggregation of many Process Steps. This kind of combination is clearly more complex and has its own limitations.
These are some different metrics categories with some examples:
- Conformance metrics (Number of deviations/variations, Protocol violations, wrong orders, etc.)
- Efficiency metrics (Number of Steps, no-show Process Steps, etc.)
- Reliability metrics (Number of Process with a wrong output / input)
- Performance metrics (Delivery time, waiting time, global duration, Number of cases not delivered on time, etc.)
- Cost metrics (Process cost, waiting cost, etc.)
- Resource metrics (Number of resources ,etc.)