Today's dataset was on tornadoes throughout the US between 1950 and 2019. After some initial exploration I noticed there seemed to be a shift in which month the number of tornadoes peaked towards earlier in the year.
I decided to group the data by decade to make it a bit easier to manage and interpret, it also highlighted trends slightly easier than looking at 70 years of data which can be quite overwhelming when put onto a chart.
The first chart I made was a number of tornadoes by month and decade heat map, and of course I had to make it a drill down to year with parameter actions. The heat map showed the shift in peak number of tornadoes well and gives the question that the dashboard focuses on.
From here I looked into the increase in the number of tornadoes for each month, this was a simple bar chart made from a couple of calculated fields to return the percentage difference between 1950 and the readers chosen year. It highlighted that those months outside of the ‘tornado season’ are increasing at a greater rate.
After seeing the dramatic increase in the winter and spring months, from my small knowledge of tornadoes I thought this might be caused by better recording equipment being able to detect lower level events. So I used a bar in bar to compare the average magnitude of the tornadoes. The findings were surprising in that the peak months had lower average magnitude that the lower frequency months.
Here's the dashboard: