Day 2 of Dashboard week is focusing on rainforests, and namely how rainforests are being damaged due to our globalising economy. The data is much more extensive than yesterdays project, and required some cleaning and pivoting in order to make it 'tableau-friendly'. Again, there was no data dictionary for this dataset, meaning that I didn't want to 'misuse' certain fields. This did limit the dataset quite a bit.
I turned to Alteryx to pivot, create a date field, and output multiple hyper files. The idea then was to use relationships in Tableau, to link each pivoted dataset to the 'main' dataset (all of the data without any pivoting required).
Establishing this relationship inside of tableau between hyper files would prevent 'exploding' the dataset (it grew from 381 to around 90,000 after joining just 2 of the pivoted datasets. After validating the figures to make sure my relationship was correctly established, I was ready to move into chart building & research.
As with most projects at the Data School, there is a lot of research required before diving into the data. Since there was no data dictionary, I turned to google to do some research. For example - what is a 'Biome'.
Ideally, I'd want to focus my entire dashboard on Boreal biomes, as they are known as 'carbon sinks'. They absorb huge amounts of carbon dioxide from the environment, but are also being heavily targeted for the high amounts of timber available that is used to burn for fuels (mostly illegally). Unfortunately only about 20 records focus on Boreal biomes, so I think widening the scope is the best bet.
I think one dashboard with filtering abilities on different Biomes, and Regions would make the most sense. That pushes the dashboard to more an exploratory kind.
With this in mind, I've decided to make this quite functionable, using some parameters and interactive titles to help the viewer know what they're looking at. A barbell chart instantly sprung to mind to look at the change in metrics such as 'Forest Loss' from the minimum available date to the maximum. I also think the general trends are going to be helpful to see, which may push the viewer to dive a bit deeper into the dashboard if they are surprised by anything they see.
Eventually things started coming together, although I am not too happy with the level of analysis I was able to get into.