This idea has been merged into another idea. To comment or vote on this idea, please visit ARISPMIYU-I-83 AI/ML features integration.
Process Mining reconstructs instances of processes that took place in the past. Both good and bad executions.
In general, the predictive tool determines future events based on the past and detect failures from defaults executed in the past.
Within the framework of Process Mining, for instances in progress, it is useful to estimate the final values of indicators.
Let's take for example the delivery time of a vehicle with a customization.
The assembly plant is known as well as the customization garage and the delivery dealer. It is therefore possible to determine an average value (estimation) for all indicators at the beginning of the process according to dimensions values (moreless like Rootcause miner).
As the process progresses, these estimate indicators can be refined with actual values (milestone crossing). A bit like a GPS that updates the initially estimated time of arrival in relation with the traffic and events that have occurred or changes in the route done by the driver.
So if the average time to arrive at the dealership is 10 days and the vehicle leaves the customization garage after 7 days depending on the activities to be carried out for this type of order, it is possible to estimate whether the vehicle will be on time or late and by how much.