Packet-level classification splits mixed packets into train, validation, and test sets based on an 8:1:1 ratio. Flow-level classification splits pcap files using 5-tuple identifiers with 3-fold validation and no overlap between sets. The dataset was uploaded by author rigcor7 to Hugging Face on October 9, 2025.
Use Cases
- Train packet classification models based on the per-packet-split methodology.
- Evaluate flow classification models using the described 3-fold validation scheme.
- Benchmark network traffic representation techniques on the defined train/val/test splits.
- Compare model performance between packet-level and flow-level classification tasks.
Strengths
- The description specifies a clear 8:1:1 split ratio for per-packet classification.
- The per-flow-split methodology ensures no intersection between train, validation, and test sets.
- The dataset was last updated on October 9, 2025, indicating recent activity.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- The dataset's source and collection method are not described.