GIS Communications - Lab 6 - Bivariate Mapping

 Greetings all!

It is absolutely crazy that this is the last module, minus final, for this class! Where did the time go? This whirlwind experience of a class is culminating with proportional symbols and one of the more unique applications of choropleth styling. That is the multivariate mapping method. In this case I am taking two different variables and visually depicting the trend between them across an area. This final week explored making meaningful proportional symbol legends, establishing an effective color scheme for bivariate choropleth mapping, and creating custom legends by manipulating graphical elements. 

The first map that I will present below is a proportional symbol map which maps the changes in the number of jobs by state. While this is only one variable being mapped, the unique thing is that there are states with positive and states with negative values. So, cumulative gains or losses, being presented. The proportional symbols provide a visual representation of the magnitude of jobs gained or loss. One tricky part of this map is that there are far larger gains than there are losses. So despite the symbol type (circles) being fairly close to one another in size, there are large differences possible on the positive side as compared to the negative side. 

For this map, I went with two separate nested legends. This is predominately because of the (as stated) incredible difference between the high value for positive growth and high value for loss. The Growth section has a much larger proportional symbol range than the loss side. Even though the smallest symbol for growth is portrayed as the same value for the largest symbol for Loss. That is a bit of an illusion when looking at the two comparatively. But in this regard the size difference is key. Additionally, the two symbols are oppositely colored. Blue for positive change, with a green background, and red for a negative change with orange background. 

The second map below focuses on comparing obesity and inactivity statistics at the county level. 

Brewer 2016, Chapter 9 expresses that multivariate mapping excels at showing the relationship and distribution of multiple related variables. In this case, I am applying a sequential – sequential arrangement to convey the interaction between these two related variables. In the previous labs, using the same data set as an example we learned that the dot plot effectively shows that the two variables are correlated. This visual depiction now shows the strength of the relationship and areal distribution of both factors combined. There are significant areas of low inactivity correlated with low obesity (so more active people with less obesity challenge). There are also significant areas of high inactivity percentages combined with high prevalence of obesity.

Unlike the previous labs where we focused on a traditional or single variable choropleth, this bivariate style provides a direct visual comparison of the spatial trends. The 3 x 3 sequential color shades give a strength guide to visualize the interaction between the two. In addition to seeing areas that are highly correlated, we can identify the areas that are particularly one sided or less directly linked, but favoring one over the other variables. Ultimately, the use of this method allows us to better understand the regional dispersion of both variables simultaneously. While this is informative, one accompanying map that I think could provide even more context would be a population distribution choropleth.

While some areas have high or low correlation, we don’t get a sense of how large or small the issue is because we aren’t seeing the total underlying population.  A large population county with a low percentage may still have more people impacted than a low population county with a high percentage problem. Thank you. 

v/r

Brandon

GIS Communication - Lab 5 - Statistics

 Greetings! 

Welcome to Stats week. We continue to work with Choropleth displays as our thematic tool of choice. However, we are adding in some more supporting details. These support statistics and information come in the form of bar graphs, dot plots, and line charts. All of this together is in effort to create an effect data visualization product. 

Before I get to the product below, I do want to discuss one of my primary challenges with this course. The amount of time for finishing. Finishing in this context are those elements of design cohesion and implementation to make an appropriately aesthetic and realized deliverable. While the product below has all of the required items, I think I could definitely enhance it with more time for finishing, or adjusting of some of the color schemes. It's not that I don't think they are effective, I simply think I could refine some more with more time. As a leader in the Air Force I often have to remind people to not let perfect be the enemy of good enough. In this case, with the time that we have, I think it is a good enough product, perhaps the 90% solution, but there is refining work I could do. I would like to introduce some post-processing outside of Arc GIS Pro that can more easily manipulate the graphical elements. 

This final layout uses a converging theme as the largest elements (the maps) on the left step down to a trend analysis on the right. The color themes of the maps and graphs help keep the focus on these items, with the background text supplementing the discussion. One improvement that I would like to have had time for would be to add shadowing or extrusion to the to both United States maps to allow those sections to establish a greater figure / ground relationship. Despite the convergence theme, I also employed an element of symmetry, though there is stair stepping involved. The two bar charts balance each other like the two maps do. Also the two types of data displays, the dot plot and line graph also balance each other with the left showing a downward trend and the right an upper trend. The amplifying text is also symmetrical about the trend line. Further, the background for this portion is a dark brown with white text to differentiate from the mapped portions. I wanted the color to stand out more, but then the text to be easily discernible when focused on. This contrast helps establish better visual balance and a greater hierarchy to the mapped and graphed areas. 

Thanks for stopping by! 

v/r

Brandon

GIS Communications Lab 4 - Choropleth Maps

 Welcome to Lab 4, discussing one of the most useful and most proliferated (in my opinion) types of thematic maps. The Choropleth map. These are a type of thematic map that use color ramps to provide valuable information about some type of enumeration unit. In the case of the below you will see an example of a couple like color ramps that I created, and one that was pulled from color brewer, a software that helps develop these color ramps for you. Then you will see a map of Colorado, with the counties as the enumeration unit. The counties are colored based on the percentage of population change from 2010 to 2014 in a diverging style based on the equal interval classification method. 

Before we get more into that lets look at some specific color ramps that I created and or worked with. So, to start, there were a set of starting values to choose from. These values are based off of an assigned value between 0 - 255 for Red, Green, and Blue. The combination of these values drives different colors, each of the individual patches in the color ramp has its own unique combination value derived from the RGB combination. 

All three of the color ramps are sequential schemes with various changes in hue, lightness, and saturation. Brewer, 2016 discusses these three things as the basis of understanding perceptual dimensions for color applications. One of the biggest differences between my linear and adjusted progression ramps to the color brewer lamp is the maintaining of the Blue-Green hue. Because the basic color from the lab given directions had slightly more blue added to the green base I kept that offset blend (ranking from most to least, green, blue, red) throughout the ramps. The sequential multi-hue color brewer ramp ends with a much lighter last step, whereas my ramps have a brighter final step. Though the linear progression and adjusted progression are very similar, the 2^nd to 3^rd and 4^th to 5^th  of the adjusted progression seem to be closer to one another than their respective steps in the linear progression. One of the biggest things that I learned after this is the adjustment of saturation to reduce the brightness of the lightest value on my two ramps. I can manipulate either the RGB number combination to get a lighter hue, or I can swap to a HSV (Hue, Saturation, Value) combination to more readily adjust singular aspects of the color, like its overall brightness. 

As a reminder, for the map below, I am working with the state of Colorado, with the data applying to the county level. The map is based on highlighting the percentage of population change over 4 years. 

 

Colorado’s population change is variable between +10% and – 13.5%. Because these highs and lows are relatively symmetric I utilized an Equal Interval approach. Additionally, looking at the histogram, the data only shows a slight positive skewness. Dealing with a somewhat normalized distribution and desire for fairly equal (3-5%) data classes on either side of 0, I decided upon the Equal Interval. One potential downside with equal interval is that it could leave some classes empty when dealing with skewed data. That is not the case here. All classes have multiple observations. I chose 7 classes because I wanted a distinctive 0 class, and each of the positive and negative classes represents approximately 3-5% of change. Now, I did look at using 5 classes, but this created too generic a distribution of class observations. That, and the range increased to approximately 3 – 7% spread. I didn’t think that breakout was as meaningful as the smaller 3-5 rundown. This method allows for a greater understanding of where the most and least change has occurred. 

Population change in this context is not good nor bad, but it is absolutely positive and negative. So I went with the Green – Purple change ramp. I think it was also complimentary to my light blue and dark gray background features. The purple and green are distinctive and have high enough hue / saturation / value to be a positive figure to the other ground features in the visual hierarchy. 

Please let me know if you have any feedback, thank you!

v/r

Brandon