mehedi107/ugr-16

UGR'16 Official Compressed CSV Archives

This public repository mirrors the official UGR'16 weekly labeled all-traffic CSV archives in their original compressed .tar.gz format.

Source: https://nesg.ugr.es/nesg-ugr16/

Organization:

• Calibration: March–June 2016
• Test: July–August 2016

The archives were downloaded inside Kaggle from the official UGR'16 download endpoint using verified HTTP byte ranges with retry/resume support. The source archives are not converted or rewritten, so their contents remain byte-for-byte identical to the verified downloaded objects. No uncompressed copy is stored.

Each archive contains the official labeled CSV flow file for its week. The attack-specific subsets, binary nfcapd files, and attack timestamp files are not included in this primary all-traffic mirror.

Example use in Kaggle:

from huggingface_hub import hf_hub_download
import tarfile

archive_path = hf_hub_download(
    repo_id="mehedi107/ugr-16",
    repo_type="dataset",
    filename="calibration/march/week3/march_week3_csv.tar.gz",
)
with tarfile.open(archive_path, "r:gz") as archive:
    csv_members = [m for m in archive.getmembers() if ".csv" in m.name.lower()]
    print(csv_members)

The metadata/ugr16_manifest.json file records source paths, sizes, and SHA-256 checksums for the mirrored archives.