Homework

1. Complete exercises 3.1d-f and 3.2c using probability plots and Q-Q plots.  Doing one plot by hand may be worthwhile, but to do the rest you should create a spreadsheet or use software.  [45 min]
2. Create a spreadsheet or computer program to compute and display "percentile plots."  (This should not be too hard: you will need to assign a unique order to each value in a batch, convert that order to a percentile, and then create an "X-Y scatterplot.")  [30-60 min].
3. Create percentile plots for some of the data distributed in Class 1 (on the last page of the handout).  Make a note of any problems you encounter.  [15 min]
4. Review the gallery of "normal probability plots" at Northwestern University Medical School. [15 min]
5. Play with the probability plot applet from the University of Alabama-Huntsville.  This web-based program lets you select two reference distributions (smooth histograms, if you will) and produces batches of data corresponding to each reference distribution.  It draws Q-Q plots of each batch.  Use large batches (n=25 is the maximum) for better results.  You can explore the effects on the plot of changing location, scale, and shape of the distributions.  [15 min]
6. Review the chapter on EDA in the Engineering Statistics Handbook.  Be aware of the context: the NIST is concerned with measurements and standards--it's one perspective out of many.  [60+ min]
7. Re-read Chapter 4.  Review this week's text notes while reading the chapter.  [1:15 hours]  Most of the exercises are worth attempting.  You can readily find software on the web to help you plot pdfs and cdfs and to generate random values.

Total estimated time: 4:00+ hours

Comments