Thread Closed 
Week 7: Ebola deaths since 1976
10-08-2014, 08:41 PM
Post: #1
Week 7: Ebola deaths since 1976
[Image: _77398785_ebola_deaths_624_v7.gif]

Source: [url=http://www.bbc.com/news/world-africa-26835233
] http://www.bbc.com/news/world-africa-26835233[/url]


Breakdown
With few attributes, the data abstraction is rather simple in this case. The items in question are outbreaks of Ebola. The attributes of this item are location (with some geo epicenter), year and number of deaths. They have created a dataset type of spatial geometry based on the location attribute.

The BBC, being a reputable news source, would suggest a task abstraction that is heavily weighted toward the analysis of this visualization by its consumers. Mostly this analysis would be for discovery -- to have people be aware of the severity of a situation. Some of the targets for this vis are the location of the outbreak, to compare the current year outbreak with previous years outbreak, to count the number of people who passed during an the outbreak and whether there were multiple outbreaks in a specific region.

The visual encoding for this graphic is straightforward. A circle mark, (the only mark that I notice) is used to indicate the location of the outbreak. Several channels are used on this mark, one of them a magnitude and the others categoric. A size channel is used on the circle mark to indicate the number of deaths. A greenish hue channel is used to indicate 2014 outbreaks while the reddish hue channel indicates 1976 - 2013 outbreaks. The luminance channel is used to indicate multiple outbreaks in a location. The location on the African map is used as a spatial global reference.

A couple of views are added to this vis. A slightly zoomed panel view on the left side of the window represents the geography of the west coast of Africa. A legend panel on the left side of the window keys the years and number of deaths.

Evaluation

Overall, this vis expresses the dataset partly accurate. The inaccuracies are information that is added through the choice of marks and channels. One such issue is that the number of deaths might be confused with the location of the deaths. The larger circles, for some viewers, might indicate the area which was effected and not the number of deaths. In fact, that was my first thought -- and I thought Ebola had engulfed western Africa! Once I read the legend, I realized that this was the number of people who have passed due to Ebola and not the area. Another issue is that the circles misrepresent the number of people. It’s tough to tell what the larger circles mean since the largest circle in the legend is much smaller than the circles on the map; there is not a representative circle for the larger ones. Consequently, it’s up to the user to scale the circles and I am still not exactly sure how a circle scales on this vis. So I am still uncertain how many people have died in west Africa due to the Ebola virus.

One point that is particularly effective from this vis is that Ebola is much more severe in 2014 than the years from 1976 to 2013. This is evident by the larger green circles compared to the smaller red ones. Also, its clear that the outbreak has shifted to the western side of Africa from the center of Africa. The overlayed circles are effective in suggesting multiple outbreaks but ineffective in where those different outbreaks occur.

Suggestions

Select a different channel for the number of deaths. Possibly luminance could be used to indicate more deaths and then match the exact range more thoroughly on the legend. Also, use a different channel for multiple outbreaks in a region. Possibly smaller constand size circles in different locations would better pinpoint multiple outbreaks.


Any other ideas? Please do share.


Side note: Ever wonder why the virus Ebola is capitalized? Click here.
Find all posts by this user
10-10-2014, 08:46 AM
Post: #2
RE: Week 7: Ebola deaths since 1976
I had the same first impression as mathewa. When I first looked at the visualization I thought that the circles represented the area effected by ebola. This isn't the case though, the circles represent the number of deaths in the region. I like the use of geography to help me locate the death toll spatially but in the epidemic regions the graphic is a bit cluttered. I can't even see Sierra Leone for example. This might be helped by using more transparent circles. That way I might still be able to see the map beneath it. I like the idea of using more luminescence to encode the change in death toll. I always find area to be a bit difficult to compare. The legend is not very helpful it only goes up to 200 deaths but the death toll in Liberia, Sierra Leone, and Guinea is clearly much higher than that. I think if they want to use area the legend should get closer to the extreme values. I like how the map has countries where there have been no cases grayed out, and those that have had at least a case of ebola are highlighted yellow. It helps me search the map quickly. Overall I find it to be an effective visualization. I get the idea though it is still a bit difficult for me to quantify.
Find all posts by this user
10-11-2014, 04:11 AM
Post: #3
RE: Week 7: Ebola deaths since 1976
I found the evaluation part of the critique interesting. The circle sizes from the legend to the map were not comparative. From my point of view, the visualization gave clear picture and supported the fact that Guinea is where the outbreak started.Although there was a part of a reverse overview+detail done here. ie Here, smaller map acts as a details and overview is the full map. Colors chosen to differentiate overview and detail were not perceptive. example both overview and detail had more blue color and thin black border. This made the detail box to look like an island in the ocean.
Find all posts by this user
10-11-2014, 11:41 AM
Post: #4
RE: Week 7: Ebola deaths since 1976
Great discussion, Matthew. You've used the theoretical language very well to analyze this viz.

This graphic is certainly timely, and I think it conveys the basic facts well. Some of the details are difficult to get, however, especially in the Gold Coast region. For example:

Choice of mark. In the main graphic the size of the bubbles and the fact that they are overlapping makes it difficult to compare them. Furthermore, not all the affected countries seem to have associated bubbles (Senegal).

Color. In the magnification on the left some details are hard to interpret. For example, the meaning of this statement is unclear: "More intense color shows multiple outbreaks." I would have thought that the reference here was to saturation, but there appear to be three different color-related differences which don't seem to be on an intensity scale: green, yellow, light blue, darker blue, red outline, and a hatched area in the NE of Sierra Leone. This is confusing. The ocean is the darker blue, clearly; unaffected regions are apparently the light blue, though nowhere is this stated. And the yellow? My intuition is that it represents affected countries, but Senegal is yellow and there is no corresponding green bubble in the main map indicating that there have been cases in Senegal. So I'm not sure about that. Additionally, the difference between the green and the yellow does not seem to be one of intensity (saturation?) but rather of hue. The hatched region? I have no idea what that represents, unless it is supposed to indicate a "less intense" color than the green, i.e., an infection rate somewhere between green and yellow. Why there is a red boundary around Sierra Leone and Liberia? Is it just to make the borders clearer? Or is it another indicator of the intensity of the outbreak? It is an unfortunate color choice, in either case, since red is already in use in the main map to indicate date range.

So, as I say, the graphic conveys the basic facts--the recent outbreak is in a different place and has a different magnitude than past outbreaks--but it is a sort of a disaster in the details. Is there Ebola in Senegal? I guess so (if yellow is indeed an "less intense" color than the green), but the graphic give us no more precise way of estimating the size of the outbreak there (no corresponding bubble) and therefore no way of comparing it with other countries.
Find all posts by this user
10-11-2014, 09:54 PM
Post: #5
RE: Week 7: Ebola deaths since 1976
It's pretty easy to parse the information given for the past ebola cases. It's not that hard to see where outbreaks happened, and about how many people dies as a result. Better detail would be nice, but it get's the point across fairly well.

Then there's the data for the outbreak for 2014. With all the green circles overlayed on each other, all I can tell is that there's a few outbreaks and a lot of died from it. It's hard to know how many people have dies so far, as it doesn't define what a much larger circle means. It looks like it means 'a lot', but I'm not sure. It's impossible to tell which circle is associated with which country, as well. Another issue I can see is the coloring of the countries on the map. I assume the tan countries are one affected by ebola at some time, and the green ones are the ones affected in 2014 but it's also not made clear, especially with Nigeria having a green dot. Having the green circles right above the legend, nestled in the corner makes the data look less important, when it's very likely anyone looking at this chart is most interested in what's happening in 2014.

There could be some colorblind issues as well, between the red and green.

So the visualization does well at showing history, but poor at what most viewers really want to know.
Find all posts by this user
10-11-2014, 11:31 PM
Post: #6
RE: Week 7: Ebola deaths since 1976
Much regarding this vis has already been pointed out. I agree that a different channel should be used to convey the number of deaths in each region. I also agree that the reference scale is lacking and as such the reader will have a hard time accurately understanding the number of deaths per region. I appreciated the focus+context technique which allowed for a part of the needed clarification.

However I am unclear as to the semantics of the channels in the focus view to understand the Ebola outbreak. The general idea is there, that significant portions of the countries have been affected however why was the change in encoding not clarified?

(10-08-2014 08:41 PM)mathewa Wrote:  [Image: _77398785_ebola_deaths_624_v7.gif]

Source: [url=http://www.bbc.com/news/world-africa-


26835233
] http://www.bbc.com/news/world-africa-26835233[/url]


Breakdown
With few attributes, the data abstraction is rather simple in this case. The items in question are outbreaks of Ebola. The attributes of this item are location (with some geo epicenter), year and number of deaths. They have created a dataset type of spatial geometry based on the location attribute.

The BBC, being a reputable news source, would suggest a task abstraction that is heavily weighted toward the analysis of this visualization by its consumers. Mostly this analysis would be for discovery -- to have people be aware of the severity of a situation. Some of the targets for this vis are the location of the outbreak, to compare the current year outbreak with previous years outbreak, to count the number of people who passed during an the outbreak and whether there were multiple outbreaks in a specific region.

The visual encoding for this graphic is straightforward. A circle mark, (the only mark that I notice) is used to indicate the location of the outbreak. Several channels are used on this mark, one of them a magnitude and the others categoric. A size channel is used on the circle mark to indicate the number of deaths. A greenish hue channel is used to indicate 2014 outbreaks while the reddish hue channel indicates 1976 - 2013 outbreaks. The luminance channel is used to indicate multiple outbreaks in a location. The location on the African map is used as a spatial global reference.

A couple of views are added to this vis. A slightly zoomed panel view on the left side of the window represents the geography of the west coast of Africa. A legend panel on the left side of the window keys the years and number of deaths.

Evaluation

Overall, this vis expresses the dataset partly accurate. The inaccuracies are information that is added through the choice of marks and channels. One such issue is that the number of deaths might be confused with the location of the deaths. The larger circles, for some viewers, might indicate the area which was effected and not the number of deaths. In fact, that was my first thought -- and I thought Ebola had engulfed western Africa! Once I read the legend, I realized that this was the number of people who have passed due to Ebola and not the area. Another issue is that the circles misrepresent the number of people. It’s tough to tell what the larger circles mean since the largest circle in the legend is much smaller than the circles on the map; there is not a representative circle for the larger ones. Consequently, it’s up to the user to scale the circles and I am still not exactly sure how a circle scales on this vis. So I am still uncertain how many people have died in west Africa due to the Ebola virus.

One point that is particularly effective from this vis is that Ebola is much more severe in 2014 than the years from 1976 to 2013. This is evident by the larger green circles compared to the smaller red ones. Also, its clear that the outbreak has shifted to the western side of Africa from the center of Africa. The overlayed circles are effective in suggesting multiple outbreaks but ineffective in where those different outbreaks occur.

Suggestions

Select a different channel for the number of deaths. Possibly luminance could be used to indicate more deaths and then match the exact range more thoroughly on the legend. Also, use a different channel for multiple outbreaks in a region. Possibly smaller constand size circles in different locations would better pinpoint multiple outbreaks.


Any other ideas? Please do share.


Side note: Ever wonder why the virus Ebola is capitalized? Click here.
Find all posts by this user
10-12-2014, 01:36 PM
Post: #7
RE: Week 7: Ebola deaths since 1976
The breakdiwn of this visualization is very good. I also agree that it can be very confusing. Once again this visualization is suffering from the effectiveness of the area encoding channel. By using the weak area channel to encode the most important data attribute it makes the whole visualization ineffective. It is hard to compare the sizes accurately between marks and it is susceptible to the relative size illusuon when the marks are clustered. Senegal is colored yellow on the map because there have been deaths there; however, since there has only been a few deaths the green circle doesn't even register as a pixel on the map.

I do agree that if the whole purpose if this visualization is to show the severity of the 2014 outbreak is far bigger than any outbreak from 1976-2013 combined, it does a good job. If the purpose is to make any real comparisons between data points, it is not an effective visualization.
Find all posts by this user
10-12-2014, 05:49 PM
Post: #8
RE: Week 7: Ebola deaths since 1976
I liked the visualization. It did a good job in highlighting the areas that are or were affected and how severely they are affected by ebola. I am a little confused by the color encoding that they have used for the mini map and they have no legend to explain that as well. I believe grey color indicates a region that is affected. The concentric cirlces for representing multiple outbreaks made the visualization a little cluttery. They could have used the size of circles to indicate the total number of deaths and number at the center to indicate the number of outbreaks. Though this would have causes a slight loss of information of deaths for each individual outbreaks.
Find all posts by this user
10-12-2014, 08:52 PM
Post: #9
RE: Week 7: Ebola deaths since 1976
First I would like to appreciate Mathew for a fantastic critique, I completely agree with the post what he has critiqued.

The visualization is completely appealing and perfect to some extent but the problem with the viz is that it should portray the actual data of how many actual deaths have happened due to ebola in the graph provided either with numbers or as a percentage of total data. The size of the bubbles doesn't make much difference.

The color of the chart should have been more eye catching and appealing.

The critique's decision to improve the viz is perfect and really appreciated
Find all posts by this user
10-12-2014, 09:52 PM
Post: #10
RE: Week 7: Ebola deaths since 1976
I would also like to thank Mathew for the great critique! I enjoyed reading it, and also agree.

In my opinion, when I first looked at this visualization I really liked it. I thought it was clear and straightforward. The legend covers everything that a user would want to know. Then I read Mathew's critique about the size of the circles of the 2014 outbreak. I understand that they are larger to show that the 2014 outbreak is worse than the cases from 1976-2013, but there should be data to go along with this. A small bar chart could be a helpful small multiple to go along with this visualization.

Also, I think the overlapping circles are confusing. Some are barely overlapping and it's hard to see what you are looking at. Perhaps the bar chart I mentioned before could be partitioned by area, city or region, and it could show the number of deaths. This could help clarify overlapping and darker colored circles as well as the number of deaths.
Find all posts by this user
Thread Closed 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)