Thread Closed 
Week 3: PCs: Not Dead Yet
09-11-2014, 11:19 AM (This post was last modified: 09-11-2014 11:32 AM by Derek Nelson.)
Post: #1
Week 3: PCs: Not Dead Yet

This visualization came from an article in the IEEE Spectrum Magazine.
Here is the direct link to the image:
The data came from a market research firm by the name of Gartner.

The designer grouped the information together and tried to show how personal computer sale growth changed over the same quarter, previous year. Using a multi-dimensional bar graph we can find the percentage of growth and the actual sales for that quarter in the years from 2009 and 2014 on this graph. The sales actually look like they are decreasing if you only look at the raw data. However, if you look at the percentage of change from the year previous it appears that people are actually beginning to purchase more pcs again. This is a quantitative and time series bar chart. That is trying to show that the pc looked like it was going to disappear but now it looks like it will be something that stays around a little longer.

I feel that this visualization is successful in trying to show that PC sales were going down, then in the last few months those sales have indeed started to grow. However, this can be hard to see. There is a lot going on with the different bars, colors, and readability of lines. The colors are really similar to the vertical bars compared to the horizontal bars. The saturation and luminance of the orange hue is really similar so it is hard to determine if the colors are different. The purple color is helping to distinguish between the vertical and horizontal bars. However, I would say that the color is chart junk and if printed in black and white you wouldn’t be able to tell between the different bars. Also the blue background was really hard to distinguish what the labels were. The text was hard to read. The size of the text was also small, which made it harder to read.

This visualization also has quantitative representation issues as well. The reference points for the axes are really hard to see where they line up on the actual bar. This is because of the 3D aspect. The pc sales bars only fall between the 60 million and 100 million sales so there is a lot of wasted space taken up by making the reference 0. Also you have no idea what the actual numbers are, it helped a little to have the value added to the bar.

There is so much information on this chart it is hard to know what to get out of it. The bubbles that give some explanation help direct to where we need to look. I really don’t see the need to have the pc sales included in this graph. A line chart that showed the growth of personal computer sales would be sufficient. However, the sale data might give us an understanding of what is actually happening.
Find all posts by this user
09-12-2014, 04:11 PM
Post: #2
RE: Week 3: PCs: Not Dead Yet
I like how you talked about the colors in the visualization. Where I agree in almost everything you mentioned. However, its clear to me that this visualization was made more to show how beautiful and cool visuals could be, more than it was made to actually show information and help the reader to come up with results in a very simple and fast way, which is the main purpose of these visualizations.

The visualization is 3D where it is hard to know the actual numbers as you mentioned and the creator tried to fix this problem by adding the numbers to the bars, but that even made it harder to follow the bar (line) since the color of the bars is the same, and my eyes were jumping from line to another trying to figure which one is which and where they end. I think if s/he made the lines closer or changed the color of the bars where two bars next to each other should have different color, (Yellow, orange, yellow, orange,yellow ..etc) would probably help more ? i'm not sure, the best way to know is actually trying it.

The visualization is absolutely eye catching and so pretty, but it didn't help much knowing the real information that it was trying to show, and so I think this visualization is not quite the best, but its not bad.
Find all posts by this user
09-12-2014, 04:19 PM
Post: #3
RE: Week 3: PCs: Not Dead Yet
Vertical bars in the chart clearly show the trend in terms of percentage change in PC sales. The 3-D perspective does obscure the actual percentage change but I think the message the author is trying to convey is only about the trend in PC sales and not about the actual values.
The actual sales values also shows the same problem associated with 3-D perspective along with an additional problem, horizontal lines merge into each other making it very difficult to follow the sales for a particular year. I think using alternate colors could have helped to separate out adjacent horizontal bars.
I agree with Derek that too much information is crammed into a single chart. A simple bar graph with actual sales figures on the bars or on top of bars or on an overlay should have sufficed as the main intention was to show that PC is not yet dead as the trend in PC sales show.

Overall, the message gets conveyed through the chart. There is confusion only when we try to look at actual sales figures.
Find all posts by this user
09-12-2014, 07:43 PM
Post: #4
RE: Week 3: PCs: Not Dead Yet
After putting on my 3D glasses, I am able to clearly see this viz. Not. Why so much 'paint'? I'll come back to this point in a moment.

First, let's talk about expressiveness. Like several of you said, the point is clear that the PC sales dropped and now are back on the rise. This is a rather simple percent change in PC sales over a six year time period. What's not clearly expressed is how the worldwide PC sales pertain to the message. The pattern of the worldwide PC sales is not the same as the PC sales change. Or maybe it's not clear the offset for the percent change in the sales. From what I can tell, the two measures are not the same (although they appear linked in the graph). Of course, I don't have the data set in front of me which might help us distinguish. Overall, I don't think the PC sales are necessary for the message.

Second, let's talk about effectiveness. Too much color. I really hope they didn't print this on paper. Color is meant to identify, group, layer, highlight. The whole thing looks like a highlighted group. Why not get it right in black and white and then highlight some of the marks? Possibly highlighting the % change bars -- with a different color for growth and decline -- might focus the viewer. Also, the blue outline of the callouts looks nice and brings attention to events in time (which are often not clearly channeled). Although we live in 2.05 viz world, I personally don't mind the 3D effect. For some reason it gives an attractive techie feel to the whole chart and doesn't lose the expressiveness.

Overall, if you can get past the atrocious tie-die explosion of color, the message is clear and the chart is layed out in a organized manner.
Find all posts by this user
09-13-2014, 08:09 AM
Post: #5
RE: Week 3: PCs: Not Dead Yet
Great graphic to choose. Derek's critique nicely identifies the key problems in this viz: too many dimensions, confusingly presented (especially with the use of 3D and color). I suppose the point of the visualization is as the title puts it: "PCs: not dead yet." That is, in a regime of declining sales the growth rate, quarter over quarter, is improving. The designers apparently feel they need these three dimensions (time, growth, sales) to convey their optimistic point. But the 3D presentation of this information, while novel, could be more simply and clearly presented in 2D, with a histogram showing sales, overlaid with a smoothed line showing rate of change (say, in red). The 3D makes it very difficult to compare bars. The bubble comment, with reference to last bar, claims that "PC shipments have begun growing again." But this fact is very difficult to discern from the heights of the bars themselves. In addition, the 3D presentation means that the growth bars on the lefthand side completely obscure the first few sales bars. So in this case the presentation actually prevents conveying the information. The colors don't help. On the lefthand side, there is an MC Escher-style effect as you try to follow the growth bar in front to the corresponding sales bar in back. Yet the yellow and magenta shade into each other, preventing this visual operation. Once again, the presentation gets in the way of the information.
Find all posts by this user
09-13-2014, 05:04 PM
Post: #6
RE: Week 3: PCs: Not Dead Yet
This particular visualization took a while for me to get my head around it. Until I had read the critique I didn't even notice that the length of the lines expanding towards the top right corner signified the worldwide PC sales. This really should have been broken down into two charts. One for the percentage growth and the other for the number of sales. As it is it is very hard to see how the bars match up, both "vertically" and "horizontally" to see an real trends. One really has to keep a very complex model in their mind just to compare the growth against the sales, particularly near the very tall bars that fade out so you can see through them. This shading causes a bit of an optical illusion and is very busy to the eye.
Find all posts by this user
09-13-2014, 05:30 PM (This post was last modified: 09-13-2014 06:16 PM by holtvg.)
Post: #7
RE: Week 3: PCs: Not Dead Yet
I think the way this visualization is trying to express the growth in personal computer sales and worldwide pc sales is clever in this 3d depiction but it takes some time for the viewer to digest what each of the dimensions means on the graph. I found it difficult to adjust spatially to the data and what parts on the 2d plane matched the 3d plane going up as you can see on the left side where it would seem as if the
data for worldwide pc sales seems to just disappear into the top out of view. The color doesn’t seem
to really help as well as I’m not sure what the random hues of purple, yellow, and orange are supposed to represent or if it’s just eye candy. Given the angled view it’s hard to quantitatively assess the data and where the z data aligns with the x and y axis data, the colors certainly do not help and I think it should just be a solid color. I would say as far as drawing the viewers attention it is very good at that but when it comes to expressing the data it’s confusing, with some bars having random data on them which isn’t helpful as well as the scale which could use a legend to clarify the units of measurement in each of the 3 axes.
Find all posts by this user
09-13-2014, 07:13 PM
Post: #8
RE: Week 3: PCs: Not Dead Yet
For this visualization, i totally agree with what derek critique, when i first saw this chart, i feel it is pretty cool, because it is colorful and 3D chart, but when i seriously see this chart, i feel that, the colors are really similar to the vertical bars and horizontal bars, and when i tried to read some information from the chart, i think i can clear see how the growth in personal computer sales changed over same quater, privous year. But it's hard to see the special value for both the growth and number of personal computer sales. And about the sales bars in the chart, i think because the value of every bars is between 60 million and 100 million, so it might make the initial mark number of sales number bar 60 million, so, it can save a lot space, and another way, the bars of sales number bars is to long to clearly see some spical month sales number, it's readability is not good enough.
Find all posts by this user
09-13-2014, 10:58 PM
Post: #9
RE: Week 3: PCs: Not Dead Yet
This one really takes me a while to know what's going on in the graph. I would like to point out several things that make me feel there is still room for improving.
1. Color: Yellow can be quite outstanding from the blue background, however, the weird pink involved in making the graph a little messy and inherent.
2. Data Density: Although it is a fancy 3D bar graph, the data density is pretty low. Lots of space has been wasted to create this 3D view. As the professor mentioned in the class, sometimes 3D will make things worse.
3. Text rotation: In order to fit the 3D view, legends were rotated. This makes this graph even hard to read. The font size of the numbers is also too small compared with this huge graph.
But I have to say this is a very unique way to combine two kinds of information together in a single graph. Although the results indicate there may be a better way, but nice try!
Find all posts by this user
09-14-2014, 04:34 PM (This post was last modified: 09-14-2014 04:42 PM by jdunstan.)
Post: #10
RE: Week 3: PCs: Not Dead Yet
M.C.Escher approves of this visualization.
I'd agree with the sentiment of the other comments.

The colors used all have a similar saturation and luminence, so the only way to see the difference is in hue. They did pick very different hues, but reducing the intensity of the background, and maybe one of the other 2 colors would be better.

It's easy to compare the height of each bar to another, and the depth of each bar to another. But it looks like there's a blurring effect when you try to track a height bar to a depth bar. (Optical illusion? I know there's a technical term for it). The design works well(enough) for putting 2 graphs that share an axis (time) into the same space, but it implies that the 2 sets of information should be compared with each other or tracked at the same time. It's easy to look at number of sales over time, or % change over time, but I can't see sales AND change at the same time.

While the use of 3D is interesting, it makes the axis labels difficult to read, particularly the x(time) axis. There are a few points where the depth bar data is obscured (or worse, made unclear) by the vertical bar. I think another set of vertical bars of another color would have worked better, in place of the depth bars.

One thing the visualization does really well is the text boxes. They describe some of the story behind the information given, which helps in understanding what I'm supposed to be understanding from the picture. But since the article itself is pretty short, the visualization is very important to conveying information. A viewer will probably look at the chart first, then read the article to figure out what the chart is supposed to represent. A better plan would be to make the visualization clearer in the first place.

I think the idea of fusing 2 charts together like this is interesting, and might have some potential. It's just not done really well here.
Find all posts by this user
Thread Closed 

Forum Jump:

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