Introduction

The data used in this analysis was sourced from the Aviation Safety Network (ASN). The ASN Database, updated every week, contains safety information and history for 15,800 airliners, military transport category aircrafts and corporate jet aircrafts since 1921. Airliners are classified as any aircraft capable of carrying at least 14 passengers.

Raw safety data since 1940 was scraped from various webpages, then cleaned and formatted to allow for analysis. This processed data is used to generate graphs and conclusions. The results from this analysis are presented with interactive graphs and relevant takeaways.

By period

First, we want to break this information down by year. Graphs below show the number of fatal airliner hull-loss accidents and fatalities per year. These figures do not take into account corporate jet and military transport accidents/hijackings.


From this visualization, we can see that airline accidents have been on a general decline since the 1940s demonstrating more rigourous safety standards and regulations. Airline casualties have also been falling, albeit since 1970.

Flight Phase

We now consider airline accidents and casualties by when in the flight phase they occured. This dimension is also considered with respect to time and includes corporate jet and military transport accidents. The flight phases are as follows: take-off (TOF), initial climb (ICL), en route (ENR), approach (APR), and landing (LDG).


The graphs above illustrate that most flight fatalities occur during the en route phase of flight travel. The second most occur during approach. This is because during approach aircrafts are close to the ground and in a more susceptible configuration.

On the other hand, while few aviation accidents occuring at high altitudes, these are the most lethal. Accidents that occur during approach or takeoff involve a slow-moving aircraft at comparatively flat angles. In contrast, aircrafts that experience failure en route fall from immense heights, sometimes reaching terminal velocity, dwarfing any real chance of surivival.

Worst geographical regions

We now consider this data by the 25 geographical regions with the most civil airliner accidents since 1945. This analysis does not include military or corporate jet accidents.



This is a choropleth map, which means that countries are shaded by the frequency of incidents. For example, the USA has the highest number of accidents, 790, and is thus the darkest color. Nigeria has the lowest, 39, and is the lighest color on the map. In second place is Russia with 485 total accidents, and Brazil in third with 185. This distribution can be partially attributed to flight frequency, far more flights are destined for and originate from the USA than compared to Nigeria.

Consider now the fatalities from these civil airliner accidents.


Again, we see the same top three countries. USA has the highest number of fatalities with 10629, Russia has the second highest with 8090, and Brazil the third with 2725. However, it is now Australia and Guinea who have the least number of fatalities

100 worst accidents

This interactive scatterplot shows the 100 worst aviation accidents, including collision fatalities and criminal occurences, but excluding ground fatalities. The data is sorted by year and measured by total fatalities. Hovering over a point provides details about the airline, country of origin, total fatalities, and year.

Summary

The chart below illustrates the total number of casualties, accidents, and survivals for each decade since the 1940s.


As we can see, fortunately both airline casualties and accidents have been consistently falling over the last half-century. This can be attributed to ever-increasing flight safety regulation and policy. Even today, aviation engineers are working on new technologies to bring these numbers even closer to zero.

While this report may be sobering, keep in mind that aviation accidents are objectively highly unlikely. A MIT professor once said that the risk of a passenger dying in an aircraft accident as 1 in 45 million, compared to the probability of a cycling fatality which the NSC estimates to be 1 in 4982.