CS 147 Homework Assignment 4

This homework assignment is due at 12 AM on Thursday, March 27, 2003 (i.e., the Wednesday/Thursday boundary). Please give your solutions to me or place them in the box outside my door.

I expect that it will take you about 4 hours to complete the assignment. Please record your actual time so I can get feedback on my estimates.

There is only one problem in this assignment: graphing data. The file prob1.txt contains observations of file sizes on a specific computer. The tab-separated fields are as follows:

Number of files observed with this size and extension.
Size observed.
Extension observed; "---" if there was no extension.

Thus, for example, the line:

	717	0	---

indicates that 717 files were observed that had no extension and a length of zero, while:

	28	35	.gif

shows us that there were 28 files that were 35 bytes long and with a .gif extension.

Prepare a graphical presentation, containing between two and four graphs, that tells a story about this data. Use the principles of graphical presentation that were discussed in class or that are given in Tufte and Jain. 80% of your grade will be based on the quality of presentation; only 20% will depend on your analysis and insights.

If you choose to work in color, be sure to print on a color printer (I can help you with that if necessary). Conversely, if you print on a grayscale printer, be sure to work only in grayscale.

To help you with your analysis, here are some things you might experiment with (individually or in combination):

Play with scales, perhaps plotting only a portion of the data.
Summarize the data only by file size (ignoring the extension) or by extension (ignoring the size).
Divide the data into buckets (making a histogram).
Look at only the number of files with a given extension, ignoring the size.
Turn one of the columns of data into a ranking.
Use kernel density estimation to smooth the data. A bit of documentation on doing density estimation in the R package is available from the University of Lyon. The R package can be accessed on Turing by typing "R" at the command line. (It's broken as of this writing, 3-5-2003, but I've created a request to have it fixed.) Documentation is available via "man R" or the R web site.

You are not limited to the above ideas. Feel free to be creative.

This page is maintained by Geoff Kuenning.