Dashboard Week Day 3 : Tornado Data

by Alex Hirst

Well, we are halfway through dashboard week. I certainly have a better understanding of where my weaknesses lie but I have also seemed to developed my time management well. Monday and Tuesday were stressful. They felt like a mad dash and frantically pulling stuff together.  Today was different. I was more relaxed and it allowed me to try some new stuff that I’d seen about on the web.

The Data : Tornado data on the US.

We’re talking starting points, ending points, magnitude, fatalities, state, damage etc. Plenty of avenues to go down. I wanted to develop the ‘why’ part of the analysis (which Andy won’t stop bugging us about).

Step 1

Step 1 is my usual. Once I know the data is clean I can go about renaming my variables so they are easier to understand at a glance.


Step 2

Now its time to analyse. For me, when I look at these type of datasets, I look for where the majority of events are occurring. Once I locate a hotspot, or an area of interest, I delve into further analysis of that spot. I went through all the natural thought processes WHERE (place) > WHAT (tornados) > How did it damage ($) > WHY did it do it. From this I could develop the spine for my analysis. Through this process I was able to develop a storyline, locating outliers, and explaining their impact with more data.

WHERE did this happen

It looks like 50% of the tornados certainly don't happen in 50% of the land. Granted it's about 30% but still, it counts.

WHAT are we talking about

Let's set the scene to see here most tornados are, so we can set some anchors in the readers mind as we go through the analysis. For example, they'll now know Texas and Kansas have way more than any other state.

HOW did it damage ($)

Now let's surprise the reader. They are expecting Texas and Kansas to have by far the highest damage costs but NO. Iowa and Georgia trump (no pun intended) the lot. The eagle-eyed readers may of seen the foreshadowing of the blue highlighted states in the first section.

WHY did this happen

But why then? Why does Iowa and Georgia have so much higher damage totals than the rest. I couldn't give them this insight and then not explain it. Firstly, severity is a possibility. Higher magnitude tornados yield more damage.

I knew this couldn't explain it completely. I had to draw in more data. I needed some idea of how Iowa and Georgia were so much worse than the rest. I looked for disaster preparedness data. I couldn't find any natural disasters which Texas and Kansas didn't also experience but I found levels of preparedness. Luckily, It said exactly what I wanted it to.

So that's it. Day 3 done. Roll on the rest!

The Design

My dashboard was inspired by one of my all-time favorites. I've waited a long time to recreate it in some way and I felt like it today. Thanks Lindsay!

Lindsay Betzendahl - Turkey Migration - Makeover Monday Week 4 2018

© 2022 The Information Lab Ltd. All rights reserved.