Business Intelligence Blog

Latest Stories

Featured Stories

Filter By Categories
Ashleigh Trinh
By
July 17, 2018

Promatix | Week 11 – Having a Blast!

After the rigours of passing my Tableau Qualified Associate exam (read last week's blog post here), this week I got to get back to playing with data and learning again. This week I have really enjoyed the topic of Makeover Monday data; VOLCANOES! I felt like I was back at school doing a science research project just this time instead of writing a long essay, I was using Tableau to make a great visualisation from the data. It was cleaner than the typical volcano science project (no bicarb or vinegar in sight) but the design and formatting capabilities of Tableau still made it fun.

The data from data.world contained just the most recent eruption date of all the volcanoes rather than the number of eruptions which was on the original visualisation, from the Smithsonian Institution, National Museum of Natural History, Global Volcanism Program database, I managed to find a list of all confirmed volcano eruptions. I decided to use Tableau Prep to clean then join the two data sets. To work with the dates which were split into individual fields for year (some of which were BCE), month number and day number, I had to convert them from numbers to string fields then I combined them to form a date in string format using a calculated field (as shown below). Then, in theory, it is easily converted to a date field by changing the data type. The issue was that Tableau currently does not have the capabilities to deal with negative date years, in fact, according to this Tableau Community Forum idea Tableau can only handle dates after 400CE. For ease, I filtered out all data prior to this date and just focussed my analyses on events from 400CE until now. My plan regarding the dates was that I could calculate the duration of recorded eruptions (if plausible), this meant that I could only do so with eruption data with a full date (i.e. dd/mm/yyyy). To overcome this since many entries had only the year then either null or 0 for the day and month, for those entries I improvised (or potentially just cheated) and set the values to 1 using ‘Group and Replace’. This meant that the date field would form, and I could use DATEDIFF.  

 

Ashleigh Blog 1

 

After cleaning the data, I joined them using a left join where the left data set was the full list of known volcanoes to-date. I feel like I am becoming more familiar with Tableau Prep now that I have used it a few times, though I am sure that there are many functions that I have yet to explore and I am excited to really see what Prep can do.

Since I enjoyed researching volcanoes so much this week, I decided to do an interactive infographic style viz to pack in lots of information about volcanoes. This was a very vague start, since there was geographic data I started with a map and worked from there. This week I decided to add some different chart types to the viz since I have a habit to just use the typical line graph, bar chart and map, this is not an issue since a lot of information is best presented with those types of visualisations, but I wanted to push the boat out this week and explore other visualisation types. It is rarely best practices to use a table to show data, especially with lots of numbers since the take home message is not necessarily easy to see instantly. However, this week I used one to show the regions of the world to use as a filter since there are 19 regions defined in the data set – shown in the map below – and are not the typical groupings (i.e. EMEA, USCA, etc or by continental) due to volcanoes also existing in the ocean and they are not restricted by land. I used the table to also show the number of eruptions per region since 400CE.

 

Ashleigh Blog 2

 

I also used a tree map to show the major rock types of the volcanoes, this was primarily because I already had a bar chart right above the space where I wanted to show the rock types. Tree maps are also great to use as a filter since the size difference will draw the user’s attention to the compartments that are vastly different in size.

 

Ashleigh Blog 3

 

Above is an image of my viz which I submitted for the Makeover Monday Viz Review. I watched the review webinar to see what feedback I got:

  • The coloured numbers in the region table makes it hard to read on the background,
    • Also, the numbers would look nicer right aligned so that the units of the numbers matched up.
  • The map is misleading when it is repeated because there are more data points than there is in the data set.
  • With this data set, the map would look great with the ‘Ring of Fire’ front-and-centre since there are an abundance of volcanoes along the Pacific Tectonic Plate. This is just cut off with a Europe centralised map so if Australia is put along the midline the Ring of Fire would be visible in its entirety.
  • The bar chart labels are difficult to read with white over the yellow and it is only half on the bar for some of the smaller bars.
  • The reason for colouring on the VEI table is unclear.
  • There needs to be better instructions regarding the interactivity of the viz.

My updated viz is embedded below after some of the changes suggested by Eva and Andy. I agreed with the alignment of the numbers in the table and the repeated map, having the numbers right-aligned makes it much easier to read. I tried to set the map so that the ‘Ring of Fire’ was in the centre however the issue after I moved the map, it meant that when I filtered to a region the map would remain at the constant zoom level and I think the map is more informative if, once filtered, would zoom in so that the viewer can see the dispersal of the volcanoes in the region. Though if there is a method where I can define the centre of the map and let the filter actions pan and zoom please do let me know, you can find me on Twitter as @Promatix_Ash. The bar chart labels are now to the left-hand side which is much easier to read, though I still think it looks nice when the bar spans the whole dashboard, but readability is more important for viewers than aesthetics. I have also underlined the instructions to guide the viewer to interact with the dashboard which will make them more visible.

 

 

If there are any other ways you think that my viz could be improved, please tweet me.

 

If you are interested in learning more about Tableau or having a free demo,

Contact Us

Subscribe Email