Reproducibility Dataset for Differential Privacy Experiments on CICIDS2017

This dataset contains the reproducibility artefacts used in the study:"An Empirical Study of Privacy–Utility Trade-offs under Repeated Differentially Private Data Releases". The artefacts are derived from the CICIDS2017 dataset, originally developed by the Canadian Institute for Cybersecurity and accessed via Kaggle:https://www.kaggle.com/datasets/chethuhn/network-intrusion-dataset The original dataset is not redistributed here. Instead, this archive includes processed and experiment-ready data representations generated as part of the study. Contents include:- Canonicalized event tables in parquet format- Temporally aggregated statistics for repeated-release experiments- Bounded datasets used for differential privacy mechanisms- Cached outputs from DP mechanisms (Laplace, tree-based methods)- Evaluation artefacts used to generate figures and results These artefacts are provided to support reproducibility of the experimental results reported in the paper. All credit for the original dataset belongs to the Canadian Institute for Cybersecurity.