This is the source code for the paper Detection of Anomal Process Behavior through Neural Networks that is currently under review for the BPM 2022.
This repository contains the following Notebooks:
- 01_log_conversion.ipynb: Converts the event logs into csv format to make it easier to load them.
- 02_data_processing.ipynb: Handles the processing, including encoding of attributes, creation of sliding windows, adding of start and end events, generation of data loaders.
- 03_anomaly.ipynb: Includes the anomaly detection algorithm, i.e. the prediction model, the loss functions and the anomaly score calculation and classification, as well as the metric computation.
- 04_training.ipynb: Includes the training phase of the neural networks for all datasets.
- 05_pdc20.ipynb: Includes the analysis for the PDC 2020 event logs.
- 06_pdc21.ipynb: Includes the analysis for the PDC 2021 event logs.
- 07_binet.ipynb: Includes the analysis for the Binet event logs, including the synthetic data sets and the BPIC datasets
- 08_latex_exp.ipynb: Covers some additional latex exports.
In order to run the notebooks, it is required to install fastai and pm4py. The easiest way to get everything running is with anaconda.
- conda create -n dapnn
- conda activate dapnn
- conda install -c pytorch -c fastai fastai=1.0.61
- pip install pm4py