It's Friday night and I'm learning rather than watching TV or wasting time on other activities that don't exercise my brain. Earlier this week, I posted my Makeover Monday viz about armed conflict in Africa using the upcoming density map feature in Tableau. As part of my plan to improve my Alteryx skills, I wanted to recreate Tableau's density maps as heat maps in Alteryx as closely as I could. In Alteryx, it all starts with the workflow:


The workflow is simple and I'm sure there's a more elegant want to create it. The guts of the workflow:

1. Create spatial points from the latitude and longitude of each record.
2. Convert the spatial points into a heat map by grouping the points into grids. 
3. Union the four streams back together.
4. Export the data as a shapefile.

The reason I have four streams in the workflow is because I didn't see a way of doing a grouping by the event type within the heatmap tool.

The Alteryx heatmaps are easy to adjust; I played with the grid size and maximum distance a few times until I got something close to Tableau's density maps (without fiddling with the settings endlessly). I settled on a grid of 33 miles and a max size of 75 miles.

I also chose to create the shapes as donuts so that they wouldn't stack on top of each other. Here's the result:

Density Maps vs. Heat Maps

Here are some of the differences I have noted:

1. Density maps in Tableau are completely dependent upon the number of marks in the view. The more accurate you want the density, the more marks you need to include. That's 9229 marks in this case.
2. Since the heat maps Alteryx generates are shapefiles, they will render much faster as there are only 28 marks.
3. The Alteryx heat maps clearly encompass all points, meaning you can see EVERYWHERE that there was an incident.
4. Tableau's density maps hide the outliers.
5. You can't have highlight actions on a density map as there no dimension categorizing the heat.
6. The density maps looks much cleaner than the heat maps.

Week 34 is a collaboration with ACLED to help them visualize armed conflict in Africa. Here's the original visualization:

What works well?

What could be improved?

What I did

The start of the 2017 NFL regular season is right around the corner, so this week we're taking a look at player behaviour, specifically, players arrested since 2000. The data come from NFLArrest.com and I used Alteryx to call their API and download and prep the data. Pretty simple workflow!


Let's have a quick look at the original viz.

What works well?

What could be improved?

My Goals

This week’s Makeover Monday is a bit of a dark topic that gives us a glimmer into the how politicians in the US respond to a tragedy. Yes, this is a single mass shooting, but they occur nearly every day in America and the responses from politicians are now all too standard. The Orlando mass shooting likely got more attention because it happened at a LGBT club. This and other tragedies beg the question: When will America get its act together?

The chart that we’re reviewing this week is by the grapics team at The Washington Post.

What works well?

• It’s neatly organised in a 3x3 grid.
• The semi-circles are all sized relative to each other.
• It’s very easy to understand.
• They used colors that are automatically associated with the two political parties
• Everything is clearly labeled.
• It’s not over-cluttered.

What could be improved?

• The chart implies that there are equal representatives in both parties. However, Republicans have 247 seats while Democrats have 187.
  
• Using semi-circles makes the shapes hard to compare.
  
• It's difficult to compare across the topics.
  
• There should be an easier way to which party talks more about each topic
  

Given these changes I would like to see, I’ve create this version that shows how many representatives talked about each topic in a tornado chart format. I then included a circle with a number inside to show which party talked about the topic more and by how many representatives. Lastly, I added lots of summary details in the tooltips.

From the Washington Post. Enough said. Let the political posturing begin.

Via the Guardian Datablog, questions to ponder:

1. Which is the most violent state?
2. Which are getting better/worse?

Click on a state either in the map or the bar chart to update the trend.  Filter to the years you want to see with the table on the lower-left.

It’s pretty clear that the peace index is worst in the south.  I wonder if there is any correlation to gun control laws?