README.md

metadata

language:
  - en
license: cc-by-4.0
size_categories:
  - 1M<n<10M
task_categories:
  - tabular-classification
tags:
  - network-intrusion-detection
  - cybersecurity
  - CICIDS2017
  - IDS
  - binary-classification
pretty_name: CICIDS2017 Network Intrusion Detection
configs:
  - config_name: temporal_3way
    data_files:
      - split: train
        path: temporal_3way/train-*
      - split: test
        path: temporal_3way/test-*
      - split: validation
        path: temporal_3way/validation-*
    default: true
  - config_name: random_3way
    data_files:
      - split: train
        path: random_3way/train-*
      - split: test
        path: random_3way/test-*
      - split: validation
        path: random_3way/validation-*
  - config_name: temporal
    data_files:
      - split: train
        path: temporal/train-*
      - split: test
        path: temporal/test-*
  - config_name: standard
    data_files:
      - split: train
        path: temporal/train-*
      - split: test
        path: temporal/test-*
  - config_name: random
    data_files:
      - split: train
        path: random/train-*
      - split: test
        path: random/test-*

CICIDS2017 Network Intrusion Detection Dataset

The CICIDS2017 dataset from the Canadian Institute for Cybersecurity, provided with temporal and random splits for fair evaluation.

Configurations

temporal (default) — Day-Based Temporal Split

Note: standard is an alias for temporal — both load the same data.

Train on Monday-Thursday, test on Friday. The model must generalize to unseen attack types (DDoS, Botnet, PortScan).

from datasets import load_dataset
ds = load_dataset("lacg030175/CICIDS2017", "temporal")  # or "standard"
# ds["train"]: 2,125,158 rows (Mon-Thu)
# ds["test"]:  702,718 rows (Friday)

Train attacks: 267,771 / 2,125,158 (12.6%) Test attacks: 288,785 / 702,718 (41.1%)

random — Stratified Random Split

80/20 stratified random split from all days combined.

ds = load_dataset("lacg030175/CICIDS2017", "random")
# ds["train"]: 2,262,300 rows
# ds["test"]:  565,576 rows

Top-20 RF Features

1. Bwd Packet Length Std
2. Destination Port
3. Packet Length Std
4. Bwd Packet Length Max
5. Avg Bwd Segment Size
6. Bwd Packet Length Mean
7. Fwd IAT Std
8. Average Packet Size
9. Packet Length Variance
10. Flow IAT Max
11. Packet Length Mean
12. Init_Win_bytes_forward
13. Idle Min
14. Idle Mean
15. Fwd IAT Max
16. Flow IAT Std
17. Flow Packets/s
18. Flow IAT Mean
19. Fwd Header Length
20. Bwd Header Length

Attack Types

Day	Attack Types
Monday	Benign only
Tuesday	FTP-Patator, SSH-Patator
Wednesday	DoS Hulk, DoS GoldenEye, DoS Slowhttptest, DoS slowloris, Heartbleed
Thursday	Web Attack (Brute Force, XSS, SQL Injection), Infiltration
Friday (test)	Bot, DDoS, PortScan

Labels

• Binary (label): 0 = BENIGN, 1 = Attack
• Multi-class (Label): 15 categories (BENIGN + 14 attack types)

Features

78 numeric flow-level features extracted by CICFlowMeter.

Preprocessing

• Removed rows with NaN/infinity values
• Stripped whitespace from column names and labels
• All features converted to numeric (float64)
• Added binary label column (0=BENIGN, 1=Attack)

Citation

@inproceedings{sharafaldin2018toward,
  title={Toward Generating a New Intrusion Detection Dataset and Intrusion Traffic Characterization},
  author={Sharafaldin, Iman and Lashkari, Arash Habibi and Ghorbani, Ali A},
  booktitle={International Conference on Information Systems Security and Privacy},
  year={2018}
}

License

CC BY 4.0 — original dataset by the Canadian Institute for Cybersecurity, University of New Brunswick.