Introduction
Air travel is gradually resuming today, and I wanted to look into airline safety, particularly the perceptions that arise through what we hear about aviation disasters. For example, some people may be less inclined to fly Malaysia Airlines due to the past news coverages of the missing Flight 370. Additionally, I am interested in comparing accidents between various airline companies, and seeing if there are airlines that are significantly more riskier than others. For this project, I will specifically look at accidents (plane crashes), both fatal and non-fatal, within a certain time period and through 56 airline companies.
Questions of Interest:
- How can aviation accidents impact business and shift passenger demand for a certain airline?
- How predictable can we be about how safe an airline is based on past statistics on incidents and fatalities?
Methods and Process
Softwares used in this lab are Tableau and Excel. The data was obtained from a free government resource through Tableau and Github. All visualizations are created through Tableau, as well as any data manipulations and developments.
I analyzed airplane crashes in several ways:
- Number of incidents based on two sets of time ranges (1985-99, and 2000-14), and airlines
- Rate of Fatalities based on airlines
- Total number of fatalities among passengers and crew by airline
To begin, I created two simple visualizations with a yellow-to-red color range to show the number of fatalities from 1985-2014. I sorted each chart so that the highest numbers (fatality count) appeared at the top. The higher the number, the more intense the color appeared in the cell. (See Figure 1 below.)
Next, I continue to compare the two time periods by creating scatter plot graphs of the fatality count. The crashes in the first graph (with the teal dots) are measured on the basis of the number of available seat kilometers (ASKs), which is defined as the number of seats multiplied by the number of kilometers the airline flies. The second graph looks at the number of incidents, both fatal and non-fatal. Each dot on the chart represents an airline, and all dots remain the same color in order to turn the viewer’s focus towards the overall pattern of the visualization. (See Figure 2 below.)
Data Development
Lastly, I tried to analyze the data in another way by creating a new calculated field from the existing dataset (see Figure 3 below). I wanted to see how many reported incidents ended up being fatal through ratios/percentages. Decimal numbers were reformatted to percentages for better readability.
I experimented with the following new visualization (see Figure 4 below) using these new calculations. I tried creating two “packed bubbles” charts where the size and color of the circles depended on the percentage number.
Results and Interpretation
In conclusion, it is tough to determine an airline’s safety reputation based on its past crash incidents. As seen in Figure 2 or in the summary image below, the first scatter plot chart shows that fatalities are difficult to predict. There is no correlation between the rate of fatalities and time period. It may be better instead to compare airlines on the basis of their number of incidents (whether or not they were fatal or not), because in the second chart (with purple dots), there is a more noticeable correlation between the two time periods. This particular scatter plot shows that some airlines obviously have more reported incidents than others.
As the years grow more recent, safety systems and airplane technologies are also likely to be improved, thus lowering the rate of accidents. For example, in Figure 1, there are evidently fewer more reported fatalities between years 2000-2014 compared to 1985-1999.
Despite these statistics covered in this lab, viewers should note that commercial airline travel is still a safe means of transit, and that the risk of being involved in a crash is very low compared to other major forms of transportation. According to PSBR, train travel, with 0.04 deaths per 100 million miles traveled, is much more dangerous than airplanes’ 0.01 deaths per 100 million miles.
Reflection
This lab overall has made me more confident in using Tableau, and has allowed me to practice developing data. One thing I could have done with the data was to create a visualization that combined the two time periods I was comparing, so that viewers could see a bigger picture of the statistics. Additionally, in trying new calculated fields with Tableau, I’m interested in other aspects of the dataset I could have worked with.