You can find part one of this blog here.
So, after the discovery of the power of Alteryx, I was left with very little time to use Tableau. Now, I had much more data than last time, but fortunately, my laptop was able to quickly load into Tableau. I decided to use a separate Tableau Workbook for this, this decision was only made for organizational reasons, and saved it accordingly. My original dashboard can be viewed here.
My original application dashboard I had grown very fond of and was reluctant to change it; my dashboard reflected how I had learnt to write academic papers during my time doing my masters, so was structured broad introduction (How does the USA compare), the topic area (Urbanisation), the subject (The USA’s MSA’s), and finally the actual question (Where do the immigrants go?). However, this was a lot of information to contain within one dashboard and was often distracting.
My intent this time was to decrease the level of exploration and make it more explanatory, so I decided to remove all but the final joined scatter plot.
The second reason for this was poor execution of some dashboard elements:
- The treemap was uninformative, these can often be uninformative due to an inability to read the smaller groups
- They also need to make up parts of a whole, which is debatable in this case as it does not represent the whole population, but it does represent the whole dataset. The main reason for removing this is to change the intent to decrease the exploratory nature and display the data in a more explanatory way.
- The top section (How does the USA Compare?) was also removed as it cluttered the page with information that was needed within the context of the dashboard, this was again to cut down the exploratory nature – there was also the removal of the following graph.
- This lollipop chart is poorly executed, the axis doesn’t start at 0, this causes the user to quickly jump to the wrong conclusions, it looks like China is way too low, but is far closer (and much closer to reality) when starting the axis at zero
The map was also taken out as it added little to the overall picture, location is not a major issue and it useful to remove it to declutter the overall message.
These removals left me with the following result (Follow this link for the interactive version).
Overall a few elements were added for increased usability:
- I continued to take New York, Los Angeles, & Chicago were removed from the main graph and given their own graph, this is a useful tool for giving your dashboard room to breath (increasing white space) and increasing the ability to draw conclusions; if they were left in the main connected scatter it would create an extremely clumped group, that would be impossible to analyse.
- I also added a key. This is to help with analysing the graph, Coach Carl pointed out during our presentations (more on this later) that while hearing me explaining it was easy to follow, for a layman’s it would be far easier to read first off with some expectations laid down.
- The data set was also whittled down from the top 50 to top 30 to new gaps opening within the data.
After these changes we were tasked with presenting them to the rest of DS13 & DS12, which is quite a nerve-racking experience. Fortunately, I have some experience in public speaking and it does come naturally, but to do it for the first time in front of your peers can still be daunting; but there are a couple of tips I’ve picked up over the years:
- Choose a few points around the room and jump between them when looking at your audience, though do this slowing, an audience can find it distracting when your head is all over the place and its never nice to stare at one point for a significant amount of time.
- Know your presentation inside out, technical issues can happen, if you know what should be coming up you will know where it is going and can sort out (or have sorted out) any issues while you present without major interruptions.
The only other major element to this presentation was doing the majority of it from Alteryx, which is quite different from using either Tableau or PowerPoint. This was quite different as there is little on screen to stimulate and a large amount of talking required to explain simple points. I would recommend that using large text blocks on Alteryx to explain each major element is a much better way of explaining Alteryx than using individual tools. For Alteryx it is even more important to understand your flow, as visual cues are less frequent and getting lost can be a lot easier without these cues.