The task:

For those who don’t know, Dashboard Week is the infamous week where we are given a dataset in the morning and create a dashboard in one day. We then present the dashboard the following morning.

For Day 1, we delved into the Global Power Plant dataset provided by the World Resource Institute.

My approach:

Along my journey at The Data School Down Under in Melbourne, I learned of the value of taking the time to understand the data and read the documentation.

However, for this challenge I did none of that.

The excitement got to me and I had formulated a story I wanted to tell before investigating the dataset. My initial idea was to visualise Japan’s use of nuclear power plants following the Fukushima Daiichi Nuclear Plant disaster in 2011.

Although the dataset contained information on powerplants, it did not contain as much as I anticipated – further inspection of the documentation revealed that database had limitations that resulted primarily in a lack of data availability.

This was in part due to lack of reporting by plants, governments and also the size of the plants.

To address this, the WRI used machine learning using the Gradient Boosted Model to predict estimates where they could. These estimates ended up being the majority of the values within the dataset and only a handful of countries had reported data shown.

However, this model only applied to certain “fuel types” such as solar, wind, hydro and natural gas. The database did not address nuclear power generation as that information was reported by a separate entity (the International Atomic Energy Agency).

I wasted a few hours trying to supplement the Global Powerplant data with data from the International Atomic Energy Agency, but ended up giving up as it was taking up too much time trying to set up the API.

Moving forward:

I lost time trying to pursue an idea when I could have spent more time towards understanding the data provided. In the end I was able to get something out of the data and a quick story (USA! USA! USA!). This included cleaning and transposing the data for my purpose and create a basic dashboard. I was even able to supplement the story with external data.

A good start. Click on the screenshot to check it out on Tableau Public

RETROSPECTIVE

What I did well
  • I have improved my reading of the documentation (It was the first thing I did) as well as looking through data.
  • Honing in on one specific subject and then changing tactic once that strategy fell through
  • Using external data sources to figure out the “why”
What I can improve on
  • Timeboxing – I believe I caught a case of paralysis by analysis and a bit of tunnel vision in in my vision
  • Finding a “story” within the data rather than trying to cater a story using a data
  • Planning ahead. Figuring out how much time to allocate before moving on

I hope you enjoyed this post.

You can check out the dashboard here.

If you would like to contact me, please feel free to connect with and message me over at LinkedIn.

Thank you for reading and have a great day!

The Data School
Author: The Data School