A1-DavidJacobs
From CS294-10 Visualization Fa07
Contents |
[edit] Good Visualization
From Alton Brown's I'm Just Here For the Food.
[edit] Explanation
This visualization is found in the appendix of an Alton Brown cookbook. The purpose of the visualization is to put some context to the various cuts of meat we often see on menus in restaurants and supermarkets. Additionally, it also provides a bit of insight as to why different cuts of meat lend themselves to different cooking methods (e.g. the cuts deriving from the head and rump are tough because the muscles are used more, implying that a wet cooking method is preferred).
[edit] Deconstruction
The data model is a set of records, one for each primal cut of meat. Each record contains a name represented as a string, it's subcuts represented as a list of strings, a preferred cooking method represented as a value 0-2 (or however you'd represent a nominal value), and finally a position within a cow probably represented as a text description in the original data.
The image model used here encodes each primal record as a region in the visualization. The size the appropriate size and shape within the cow. The color of region denotes which cooking method is preferred, while the text labels the primal along with it's list of subcuts.
[edit] Critique
The visulization is simple and informative. As a datamap, the chart not only allows for understanding subject matter at a glance, but also makes it possible to display the spatial relationships of the various primals. Specific cuts of meat (ones that you would recognize on a restaurant menu) are grouped under the primals they are derived from. Additionally, color is used to indicate preferred cooking methods and actually can be used to draw conclusions about the quality of meat in relation to cut position.
[edit] Bad Visualization
From Maurice H. Holstead's Elements of Software Science.
[edit] Explanation
This chart shows the results of an experiment testing the validity of an equation predicting implementation times for particular programming tasks. It's purpose is to demonstrate the accuracy of the theoretical formula. Direct comparisons can be made between adjacent columns, and overall accuracy can be gleamed from the statistical summaries provided along the periphery.
[edit] Deconstruction
Each programming task has a record in which there are three observed times represented as integers and three predicted times also represented as integers. This is encoded visually as a simple table, grouping conceptually similar numbers spatially. The visualization also provides some statistical analyses (since difficult to gleam from the numbers themselves), though technically redundant given the raw data.
[edit] Critique
Although the chart does a better job of presenting the data than a raw list of results (a few pages earlier in the text), it still is difficult to grasp a whole lot of meaning beyond the statistics provided. A scatter plot of observed implementation time vs. predicted implementation time would more quickly reveal the nature of this distribution. Any points far off the diagonal would represent errors in prediction and be easily identifiable.
[edit] Redesign
In my redesign, I've created a relational scatter plot showing actual programming time vs. theoretical programming time. Since there were three trials for each task (with differing theoretical values), I plot each trial in a different color. This way, it's possible to see if the formula is particularly weak for earlier or later implementations. Additionally, I've plotted a diagonal line to make visual comparisons between data points and the ideal line.
