Thread Closed 
Week 10: Fern Fossils
10-30-2014, 10:07 PM
Post: #1
Week 10: Fern Fossils
This visualization shows data gathered from a exceptionally well preserved fern fossil.

http://blogs.scientificamerican.com/artf...l_2014.jpg

Full article, which has some further description of the visualization:
http://blogs.scientificamerican.com/artf...ng-detail/



The visualization is split into 3 views:
1) A simple 'legend' view with the color code and image for a living and fossilized fern cell
2) A box and whiskers plot showing the distribution of perimeter and area of cell nuclei
3) A chart comparing nuclear perimeter and area

The first view is quite simple and straightforward, simply indicating that blue is for living cell data and brown is for fossil cell data.

The second view encodes data spatially along the vertical axis, and uses color and horizontal spacing for similar sets of data.It shows interquartile ranges, mean, median and extremes for nuclear perimeter and area for each cell type.

The third view is the most complicated. In addition to spacial and color encoding, it also uses shape to encode if a point on the chart is a minimum or maximum diameter for that perimeter. It also has a line for best linear fit for the max and mins of each cell type.

Considering that the point of the visualization and article is to show how similar prehistoric and modern fern plants are, I think it does that job very well. In the second view we see that the averages and extremes for the cell types are very close to each other, and in the third one we see that they follow a similar pattern of growth.
There is a lot of overlapping in the third view that obscures some of the details, but details are not the point of this visualization and the general trend is not obscured.
Find all posts by this user
11-01-2014, 08:52 PM
Post: #2
RE: Week 10: Fern Fossils
I agree with jdunstan's critique. The visualization does efficiently communicate the similarity between modern and prehistoric fern plants. I think that the colors that were chosen are a little dull. I think they should have used brighter colors to help increase memorability and readability. It may also have been helpful to highlight the nuclear perimeter and the nuclear area in the images. I also think that using arrows in the graph is a little much. If the arrows are close together you can't really tell which is which, and if they are far apart, you can infer that information for yourself. Overall I think this was an effective visualization.
Find all posts by this user
11-02-2014, 10:48 AM
Post: #3
RE: Week 10: Fern Fossils
This is a traditional vis for the scientific paper. Unlike these fancy vis we've seen a lot previously, this one is more like the vis we always see when we are reading papers, and more likely we will create such figures in our own thesis or paper in the future. Therefore, this is definitely a good one to do a critique.

When creating a figure for a paper, the first thing needed to be considered is what you want to show in this figure and how many rows or columns are in this figure. In this case, the writer wanted to show how similar prehistoric and modern fern plants are by showing a figure with two rows: the first row shows the nuclear perimeter and area for prehistoric and modern fern are similar and the second row shows he nuclear diameter are similar using a scatter plot. There are three graphs in this figure, but they all present the same topic.
The main issue in this vis: color encoding and shape encoding of the categorical data in the scatter plot?
As we've learned in the class, when representing categorical data, color hue and shape are two powerful ways. However, in this case, the writer used regular triangle and inverted triangle to represent max and min. I guess the writer wanted to show the "max" and "min" meaning by using these triangles. (eg, inverted triangle shows all the values are smaller than the max value, thus representing the max). However, I don't think it's a good way in this case. Since the regular triangle and inverted triangle are too similar, it's very hard for readers to tell which one is which in this graph. Also, overlaying effect are strong and covered some information.
Find all posts by this user
11-02-2014, 09:06 PM
Post: #4
RE: Week 10: Fern Fossils
This seems like a very simple visualization.
It is quite easy to interpret.
The only critique I have with this visualization is that the shapes and colors used as encoding channels do not seem to convey the information very effectively. Although it was a clever idea to use a triangle for max and an inverted triangle for min, when they all cluster like in this graph, it becomes difficult to read. The overlapping min and max forming the star shape almost go unnoticed.
Other than that, I find the visualization quite good.
Find all posts by this user
11-02-2014, 11:36 PM
Post: #5
RE: Week 10: Fern Fossils
This is very simple visualization but quite an effective one.

Pros---
1)Color and spatial encoding is used to differentiate between living and fossilized fern cell.
2)Use of box plot in second visualization is a good idea as it shows different statistical information like interquartile ranges, mean, median and extremes for nuclear perimeter and area for each cell type.

Cons---
1)In the third figure use of arrows to show maximum and minimum is not a good idea as it is difficult to look for overlapped ones.Also when the brown arrows of similar orientation overlaps then the brown one will hide the blue one completely.
Instead squares and dots could have been used to make it a more effective visualization.
Find all posts by this user
11-03-2014, 03:07 AM
Post: #6
RE: Week 10: Fern Fossils
A few quibbles:

1) For the graphs on the top right, do they include outliers in the range? It doesn't seem so from comparing to the scatterplot.

2) The rate of growth of perimeter vs diameter seems to be about pi (which is what one expects for circles from standard formulae). It would have been nice to see an illustration of this on the graph, perhaps by showing a y = pi x type line.

3) As we've discussed in some of the lectures, conjunctive visual encoding - e.g. by shape + color simultaneously - is not very distinctive and does not "pop" out. With only 4 items to track, perhaps just encoding by color would have been clearer.

4) The grid lines seem extraneous, especially since as commented earlier in the thread the point here is not really detail, and the symbols overlap too much to allow much precision.
Find all posts by this user
11-03-2014, 11:00 AM
Post: #7
RE: Week 10: Fern Fossils
I would agree with your critique. It really does show that there is a similarity between the two cells types. At first it was not that clear as to what but after some description it is clear that it is similar. I agree that the legend is giving the rest of the chart context but I am not so sure what the picture of the cell gives the graph. Otherwise this was a very well done visualization.
Find all posts by this user
Thread Closed 


Forum Jump:


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