Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/miraculixx/flightdelays

US flight delays
https://github.com/miraculixx/flightdelays

Last synced: 24 days ago
JSON representation

US flight delays

Awesome Lists containing this project

README 

        



  

  

  

  

  



  


    


      
Effective Data Visualization


by Patrick Senti, [email protected]


View at https://miraculixx.github.io/flightdelays/index.html


Summary


Flight delays are a cause of dissatisfaction for most if not all airline passengers. This story looks

at the cause of flight delays using data provided by the US Bureau of Transportation Statistics,

https://www.bts.gov/ throughout the years 2003 - 2017. The story shows that just 3 carriers have caused 

1/3 of all delays, and that most delays are caused by the airlines and the national aviation system (NAS)

and operating in highly frequented regions across the US. Surprisingly, the often suspected extreme weather 

conditions and security checks at airports cause less than 5% of all delays.


Design


My design choices after a few iterations (see below) were as follows:


    

  1. 

    Use a line plot to show change of the causes (in percent) over time. I tried a few other 

       options like box plots and bar charts, however they were hard to read and did really convey

       the message (in particular, that some causes have increased or declined where others are

       stable or even negligible). The line plot includes a color key as a legend.
    

    2. 

  2. 

    Use a bar chart to show cause of delays by carrier, overall. I tried a pie chart but found

       it hard to interpret. The bar chart is clean and conveys the information very well.  
    

    3. 

  3. 

    Use a map to show where (in which regions) the most delays occur. I found a map to be a naturual

       way to depict this information, since the data is aggregated by geographical region. The downside

       is that the map does not allow a quick realization of actual figures, so I added a few interesting

       data points in the explanatory text next to the plot.
    

    4. 



Following feedback received by the kind people in the Udacity formus, I have adjusted the design as follows:


    

  • 

    Variables are now labeled properly, in plain English (instead of their technical names)
    

    • 

  • 

    All plots' axis grids are reduced to the minimum, in particular the line plot does not have any

      grid lines, the bar chart only shows horizontal grid lines. This improves readability and is really

      pleasing for the eye.
    

    • 

  • 

    The line plots's stroke width was increase to make the colors more easily visible and comparable to the

      legend.
    

    • 



Prior to publishing my visualization, I already tried out several ideas:


    

  • 

    Adding a "year selector" to each graph, so that only particular years would be shown. This proved

      to be quite complex and destracting, so I removed it again.
    

    • 

  • 

    Using a box plot on percentage of delayed flights, by airport and year. This was an interesting plot

      during data exploration, but it proved too complex to show in the final report. In particular there are

      so many outliers that one can hardly see anything on the plot.
    

    • 



Feedback


Details see https://github.com/miraculixx/flightdelays/issues


    

  1. 

    Nice charts, I am able to understand them all, because I was working on the flight delay dataset too.

       Other viewer may have label problem to understand.

       "Arr_Del15” is better to change to “Arrived Delay”(or something else) with a percentage value in the second chart. 

       (comment by fxietech)
    

    2. 

  2. 

    for the first graph, I noticed the coloring for the line graph was hard to distinguish, when referring to the legend. This issue was resolved when the user hovered over the line.

       (comment by Raul M.)
    

    3. 

  3. 

    Coming to the second graph, I noticed the gridline in the background. It didn’t truly distract me from the graph, but it can be a cause of distraction for visualization, per lesson plan.

       (comment by Raul M.)
    

    4. 

  4. 

    I enjoyed reading your project. The format and documented transitions were appealing for an exploratory data analysis project.

       The third graph looked good! I didn’t feel there were any necessary alterations for it.

       (comment by Raul M.)
    

    5. 



Resources


    

  • Data source: Monthly US Flight Cause Delays Bureau of Transportation Statistics, Flight cause delays, 2003 - 2017,

      https://www.transtats.bts.gov/OT_Delay/ot_delaycause1.asp?type=3&pn=1
    • 

  • Include Github comments on the page, https://stackoverflow.com/a/26608674
    • 

  • Hosting web pages on github, https://help.github.com/articles/user-organization-and-project-pages/
    • 

  • Github comments web page integration, https://stackoverflow.com/a/26608674
    • 

  • 

    3rd Party Libraries used under respective license: 
    

      

    • d3 - https://d3js.org/
      • 

    • d3plus - https://d3plus.org/
      • 

    • d3-geomap -https://d3-geomap.github.io/
      • 

    • topojson - https://github.com/topojson/topojson
      • 

    • bootstrap - https://github.com/topojson/topojson
      • 

    • backbonejs - http://backbonejs.org/
      • 

    • underscore -  http://underscorejs.org/
      • 

    • jquery - https://jquery.com/
      • 

    • requirejs - http://requirejs.org/
      • 

    

    • 



Implementation


Code


The app is implemented in a modularized approach using requirejs AMD modules. Backbone.js is used to implement

the plots (as Views). In this model, the index.html provides the page layout, app.js implements application

logic by instantiating views as specified (declarative), and the views modules implement and encapsulate all

the d3 and d3plus logic. This separation of concerns has many advantages, in particular the app is easier to

extend and change. In addition, most of the code is independent of the specific data and could be easily 

adopted to a different data set and a different data story.


Evidence of improvements


    

  • Initial version - see initial_version.pdf or commit https://github.com/miraculixx/flightdelays/tree/initial 
    • 

  • Published version - see published_version.pdf or commit https://github.com/miraculixx/flightdelays/tree/published
    • 

  • Final version - see https://miraculixx.github.io/flightdelays/index.html or https://github.com/miraculixx/flightdelays/tree/master
    • 



Data files


    

  • 

    File with base data: 
    

      

    • https://github.com/miraculixx/flightdelays/blob/master/raw/fligth_delay_cause_monthly_2003_2017.csv
      • 

    • as downloaded from https://www.transtats.bts.gov/OT_Delay/ot_delaycause1.asp?type=3&pn=1
      • 

    

    • 

  • 

    Files used in visualization
    

      

    • Cause of delays: https://github.com/miraculixx/flightdelays/blob/master/data/delays_cause_year.csv
      • 

    • Cause by carrier: https://github.com/miraculixx/flightdelays/blob/master/data/causeby_carrier.csv
      • 

    • Cause by airport: https://github.com/miraculixx/flightdelays/blob/master/data/airport_causes.csv 
      • 

    

    •