Package for computing the largest eigenvalue based measure in the context of process mining and regular languages. Description of the technique is provided in the paper Behavioural Quotients for Precision and Recall in Process Mining.
You'll need to install lpsolve55 and gradle.
For example in ubuntu run in terminal:
$ sudo apt-get install lp-solve gradle
Within Linux-like systems, you need to:
- first get the sources:
git clone https://github.com/andreas-solti/eigen-measure.git - Run the shell script to download the dependencies (due to licensing, I did not package them automatically)
$ ./download_dependencies.sh
Install Ubuntu for Windows and follow steps above from within the virual linux environment.
You can verify, if all works as expected by running:
gradle test
Last, explore the sources (src) in your favorite editor (e.g. IntelliJ, VS-Code, or Eclipse). There are some tests that show the functionality. Unit tests can also be run from inside those editors.
The class to use for computing precision and recall is
org.processmining.eigenvalue.automata.PrecisionRecallComputer
It uses an AcceptingPetriNet as input and an Event log with the classifier to compute the PrecisionAndRecall.