A2-AmandaAlvarez
From CS294-10 Visualization Fa07
The data I used is from the National Transportation Safety Board's Annual Review of Air Carrier Accident Data for 1998. The data covers the period 1983-1999 *, and lists the dates and locations of accidents, as well as aircraft damage, fatalities and injuries, causes of crashes, the stage of flight at which the crash occurred, etc. The Excel spreadsheet of the data can be found here [1]. I used Tableau to create these visualizations.
( * I don't know either why the report for 1998 contains the following year's data.)
The original spreadsheet did not require any transformations to be imported into Tableau. I summed up all the fatalities, which in the data were divided into six categories (crew, passengers, others...). The data consisted of more categorical dimensions than quantitative measures, and so wasn't the best for answering very meaty questions, but it had some interesting information. Starting with some elementary questions about the data...
[edit] Q1: How do numbers of accidents and fatalities progress over time?
Number of accidents remains fairly constant, while the fatalities fluctuate wildly.
[edit] Q2: Cause of accidents over time?
In the data each accident is assigned a broad cause: environment, personnel, or aircraft. The second plot shows how the numbers of accidents attributed to each of these factors progressed from 1983 to 1999. There is a lot of fluctuation in all three factors, but the dashed trend lines show that the numbers of accidents due to aircraft and environmental factors was quite constant over sixteen years, whereas the number of accidents caused by personnel decreased (though personnel remained the main cause of accidents over the whole time period).
[edit] Q3: At what stage of the flight did accidents happen?
I looked at different times during the flight when accidents occurred eg. takeoff, landing, cruising, etc. Presumably the crew is more tired at the end of the flight (approach and landing), so I expected to find personnel contributions to accidents to spike at those stages. It turns out personnel factors don't contribute to accidents at the end of the flight (proportions between accident factors remain constant throughout phases of flight), though the end of the flight is when the most accidents occur.
Flight stages progress from STANDING to TAXI and so forth, with the end being LANDING
- - - - progression of flight - - - ->>
[edit] Q4: Which engine type contributes to accidents at each flight stage?
Reciprocating engines are the cause of more accidents than other engine types; these accidents are less serious though, not leading to many fatalities. There are more accidents during landing but more fatalities in the accidents that occur during takeoff. Reciprocating engines cause more accidents, but accidents caused by turbofan engines result in more fatalities. Turbojet engines are the safest type throughout the flight.
[edit] Q5: Effect of weather?
In plot 1 accidents remained constant over 16 years, and here in plot 5 we see that accidents occur uniformly throughout the year, despite the flight weather conditions being worse in the winter and better in the summer (red bars indicating poor weather are larger in the winter and smaller in summer). IMC weather conditions are worse than VMC conditions. There is a sharp peak in fatalities in December.
[edit] $64,000 Question: WHEN IS IT SAFE TO FLY?
The weekend. The least accidents occurred during the weekend for the 16 years that this data covers. Also, there are fewer fatal accidents on the weekend. Since it is safe to say that there are more passengers on the weekends, and accidents appear to go down on the weekends too, it really is safer to fly on the weekend. The above plots indicate that accident frequency has remained relatively constant (it has not become safer to fly than in the past), but accidents caused by personnel have decreased (better safety training?). It seems that certain phases of the flight and particular engines contribute to accident likelihood (a bad combination can increase the chance of an accident), and that flying when others fly (the weekends) may decrease the chance of an accident (though perhaps not in December).
