README.md

metadata

configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  features:
    - name: id
      dtype: int64
    - name: aircraft
      dtype: string
    - name: operator
      dtype: string
    - name: registration
      dtype: string
    - name: flight_phase
      dtype: string
    - name: flight_type
      dtype: string
    - name: survivors
      dtype: int64
    - name: crash_site
      dtype: string
    - name: schedule
      dtype: string
    - name: msn
      dtype: string
    - name: yom
      dtype: int64
    - name: crash_location
      dtype: string
    - name: country
      dtype: string
    - name: region
      dtype: string
    - name: crew_on_board
      dtype: int64
    - name: crew_fatalities
      dtype: int64
    - name: passengers_on_board
      dtype: int64
    - name: passengers_fatalities
      dtype: int64
    - name: other_fatalities
      dtype: int64
    - name: total_fatalities
      dtype: int64
    - name: circumstances
      dtype: string
    - name: crash_cause
      dtype: string
    - name: timestamp
      dtype: string
    - name: latitude
      dtype: float64
    - name: longitude
      dtype: float64
    - name: image_url
      dtype: string
    - name: hex_code
      dtype: string
    - name: external_id
      dtype: string
  splits:
    - name: train
      num_bytes: 10260424
      num_examples: 27598
  download_size: 10260424
  dataset_size: 10260424
pretty_name: PlaneCrashData - Global Aviation Safety & Accident Database (1918-Present)
license: cc-by-4.0
tags:
  - aviation
  - transportation
  - flight-safety
  - plane-crashes
  - aircraft-accidents
  - aviation-history
  - geospatial
  - time-series
  - open-data
task_categories:
  - tabular-classification
  - tabular-regression
  - text-classification
  - time-series-forecasting
size_categories:
  - 10K<n<100K
language:
  - en

PlaneCrashData — Global Aviation Safety & Accident Database (1918–Present)

PlaneCrashData is a comprehensive, open historical dataset cataloging global aviation accidents and incidents covering over a century of civil and military aviation from 1918 to the present day.

The dataset aggregates 27,598 verified aviation accident records with rich structural metadata, including aircraft technical specifications, operator/airline records, flight phases, full textual circumstance narratives, causal categorizations, survival and fatality statistics, and geospatial coordinates.

Accompanying interactive web visualization: www.planecrashdata.com.

---

Key Highlights & Dataset Statistics

• Total Recorded Incidents: 27,598
• Temporal Coverage: May 1918 – August 2026 (over 108 years)
• Unique Aircraft Types/Models: 1,187
• Unique Airlines & Operators: 9,002
• Countries & Territories: 226
• Total Fatalities Recorded: 151,390
• Geospatially Geocoded Records: Coordinates (latitude, longitude) for geospatial GIS analysis

---

Dataset Schema & Fields

Field Name	Type	Description	Example
id	int64	Unique incident identifier	1
aircraft	string	Aircraft model / type	Boeing 737-800, Douglas DC-3
operator	string	Airline or operating organization	Pan American World Airways, USAF
registration	string	Aircraft registration / tail number	N737AA, G-ABCD
flight_phase	string	Phase of flight when incident occurred	Takeoff (climb), Cruise, Approach, Landing
flight_type	string	Category of flight operation	Scheduled Revenue Flight, Cargo, Military, Test
survivors	int64	Number of verified survivors (nullable)	15
crash_site	string	Physical surroundings / environment	Airport (less than 10 km from airport), Mountains, Sea
schedule	string	Origin and destination route	New York - London, Paris - Rome
msn	string	Manufacturer Serial Number (MSN / c/n)	24680
yom	int64	Year of aircraft manufacture	1998
crash_location	string	Detailed geographical location description	Tenerife-Los Rodeos Airport Canary Islands
country	string	Country of incident	United States of America, Spain
region	string	Geographical region / continent	North America, Europe, Asia
crew_on_board	int64	Number of crew members on board	8
crew_fatalities	int64	Crew fatalities count	0
passengers_on_board	int64	Number of passengers on board	150
passengers_fatalities	int64	Passenger fatalities count	0
other_fatalities	int64	Ground / non-occupant fatalities count	0
total_fatalities	int64	Total fatalities (crew + passengers + ground)	0
circumstances	string	Detailed narrative description of the event	Narrative text...
crash_cause	string	Categorized primary cause	Technical failure, Pilot error, Weather
timestamp	string	UTC timestamp in ISO-8601 format	1977-03-27 17:06:50+00:00
latitude	float64	Latitude coordinate in decimal degrees	28.4827
longitude	float64	Longitude coordinate in decimal degrees	-16.3415
image_url	string	Representative or historical aircraft image URL	https://...
hex_code	string	ICAO 24-bit Mode S transponder hex code	4005B5
external_id	string	External database reference ID (e.g. ASN)	19770327-0

---

Quickstart & Usage

Using Hugging Face datasets (Python)

from datasets import load_dataset

# Load dataset
dataset = load_dataset("MauroPello/planecrashdata")

# Inspect dataset
print(dataset["train"][0])

Using pandas or polars directly with Parquet

import pandas as pd

# Load Parquet directly from Hugging Face Hub
url = "hf://datasets/MauroPello/planecrashdata/data/train-00000-of-00001.parquet"
df = pd.read_parquet(url)

print(f"Loaded {len(df):,} incidents spanning {df['yom'].min()} to {df['yom'].max()}")

import polars as pl

# Fast Polars scan
df = pl.read_parquet("hf://datasets/MauroPello/planecrashdata/data/train-00000-of-00001.parquet")
print(df.group_by("region").agg(pl.col("total_fatalities").sum()))

Using DuckDB (SQL)

INSTALL httpfs;
LOAD httpfs;

SELECT 
    country, 
    COUNT(*) AS total_crashes, 
    SUM(total_fatalities) AS total_fatalities
FROM 'https://huggingface.co/datasets/MauroPello/planecrashdata/resolve/main/data/train-00000-of-00001.parquet'
GROUP BY country
ORDER BY total_crashes DESC
LIMIT 10;

---

Applications & Use Cases

1. Aviation Safety Analysis: Long-term safety trend evaluations, flight phase vulnerability analysis, and accident rate comparisons across decades.
2. Predictive Modeling & NLP: Cause classification and risk factor extraction from historical accident narrative circumstances.
3. Geospatial & Temporal GIS: Mapping aviation incident hotspots, terrain correlation, and regional safety improvements.
4. Historical & Academic Research: Cross-referencing fleet evolution, aircraft manufacturer safety records, and regulatory milestones.

---

License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

---

Citation

@misc{planecrashdata2026,
  author = {Mauro Pellonara},
  title = {PlaneCrashData: Global Aviation Safety and Accident Database (1918-Present)},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/MauroPello/planecrashdata}}
}