A Two-Stage Predictive Process Monitoring Approach with Enabled State Filtering for Variable Suffix Scenarios

Installation

1. Create a python environment

   conda create -n ESF python=3.8.0
   conda activate ESF 

2. Install pytorch

   Following the official website's guidance (https://pytorch.org/get-started/locally/), install the corresponding PyTorch version based on your CUDA version. For our experiments, we use torch 1.12.1+cu116. The installation command is as follows:

   pip install torch==1.12.1+cu116 torchvision==0.13.1+cu116 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu116

3. Install other related dependencies

   pip install -r requirements.txt

Train

First, you need to specify data_path and dataset in configs/ESF_Model.yaml.

Here, Two training methods are provided here:

1. Specify Hyperparameters: Specify model_parameters in configs/ESF_Model.yaml.

   python execute/ESF/train_ESF.py

2. Use Optuna for Hyperparameter Optimization:

   python execute/ESF/run_ESF.py

Test

Evaluate the model:

python execute/ESF/test_ESF.py

If you the code from this repository, please cite our paper:

@inproceedings{Chao Zheng,
  title={A Two-Stage Predictive Process Monitoring Approach with Enabled State Filtering for Variable Suffix Scenarios},
  author={Chao Zheng, Jian Wang, Bing Li, Chao Zheng},
  booktitle={2025 IEEE International Conference on Web Services (ICWS)},
  year={2025},
  pages={},
  publisher={IEEE}
}