Data-based Process Variant Analysis

This repository contains the python implementation of the Data-based Process Variant Analysis paper.

Setup

To run the tool, you need to install a custom version of ipyevents (which is located in this repository as well).

Install the python requirements of this tool, like Jupyter Labs, PM4Py first. Then install the custom ipyevents package locally by

$ cd custom-ipyevents
$ pip install -e .
$ jupyter nbextension install --py --symlink --sys-prefix ipyevents
$ jupyter nbextension enable --py --sys-prefix ipyevents
$ npm install
$ npm run build
$ jupyter labextension install

Usage

To reproduce the findings of the paper, you can download the MIMIC-IV dataset on your own and generate an event log to feed into the tool. For doing so, you can use the Juypter notebook located in notebooks/DBPVA_Event_Log_Generation.ipynb.

With the generated event logs, or your own event logs, you can then use the VariantComparator.

The most simple way to use the tool is by only using the VisualVariantComparator. In notebooks/Showcase.ipynbthere are some examples on how to use the package for variant comparison.

To launch the tool, the following steps are required:

1. Read your event log
2. Split the event log by an arbitrary criterion
3. Initialize the Variant Comparator by

variant_comparator = VariantComparator(split_log_1, split_log_2, full_event_log, 'Name Split 1', 'Name Split 2')
variant_comparator.prepare()

4. Start the VisualVariantComparator by

visual_variant_comparator = VisualVariantComparator(variant_comparator)
visual_variant_comparator.show()

If you only need specific parts of the variant comparator, have a look in the vadbp/variant_comparator.py file.