Process Mining - Event Logs Replay
Event logs replay is a useful tool to visualize the history of events. A case is represented by a token in the graph. If an event occurs, a token is then placed on the corresponding node in the graph. An edge between two nodes represents the progress between two events.
To replay the event logs we need two things:
- Event logs
- Model: This is basically a dependency graph.
There are techniques to build a model from event logs. We will cover this topic in other posts. In this post we would like to focus on the visualization part so we first define a model manually and then use it to generate an event log. You may download the generated event logs here.
Some notes on the visualization:
- The edge represents the progress of moving to next event so the length of edges doesn't mean anything.
- As the length of edges doesn't mean anything, the speed of tokens doesn't mean anything either, which is very unfortunate.
- What matters is the time a token spends on a edge.
- The graph does show the "spatial" distribution of tokens. At a given time, the graph represents a snapshot of the progress of all cases.
- A node has input flow and output flow. If it takes time for next event to occur, we will see tokens accumulated on the outgoing edges. For example, in the demo below we will see there are congestion between d and f.
- Tokens should start from start state and end at end state.
- The average latency between two events can be represented by the stroke width of the corresponding edges.
----- END -----
If you have questions about this post, you could find me on Discord.
©2019 - 2023 all rights reserved