Cabin Safety Incident Risk Levels for 8,280 Global Reports from 2004–2024

8,280 global cabin abnormal event reports from 2004 to 2024 were used to train a hybrid CNN-LSTM-Attention model for risk classification. The model achieved 95.01% accuracy and an F1 score of 94.17% on its test set. The dataset was authored by Lianbin Zhou and shared on figshare under a CC-BY-4.0 license.

Use Cases

• Benchmarking machine learning models for text classification based on the reported 95.01% accuracy of the CNN-LSTM-Attention model.
• Developing intelligent early warning systems for cabin safety risks based on the described 'text feature–risk mechanism-hazard level' mapping.
• Studying patterns in global aviation incident narratives for data-driven decision-making in safety management.
• Training models to extract deep semantic features from safety reports as demonstrated by the proposed hybrid model.

Strengths

• Model performance is quantified with specific metrics: 95.01% accuracy and 94.17% F1 score on the test set.
• Data volume is explicitly stated as 8,280 incident reports.
• Temporal coverage is clearly defined, spanning 20 years from 2004 to 2024.
• The framework establishes an interpretable 'text feature–risk mechanism-hazard level' mapping system.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the performance data is unknown, which may limit suitability assessment.
• The dataset is very small at 5.5 KB, indicating limited scope, likely containing only summary statistics.

Provenance

Source

figshare, authored by Lianbin Zhou.

Collection Method

Likely derived from the analysis of 8,280 global cabin abnormal event reports using a deep text mining and intelligent risk assessment framework.

Time Range

2004 to 2024

Freshness

Last updated 2026-05-27 17:40:01; freshness should be verified.

Geography

Global