Assignment 1. Due date September 30th.

Question 1. Task: Exploratory data analysis of the car89 data set.
Car89.txt Tab delimited, missing values denoted as "NA".
Car89.jmp

Description of variables

This assignment will give you the opportunity to use the exploratory data analysis tools that we have discussed in class.

The objective is to write a report of about three pages in length that describes the main features in the data set. The report should be accompanied by a set of graphics which are referred to and described in the report itself. You should prioritorize your findings and present them in a structured manner.

The exploratory stage is the first stage of any analysis, but the ultimate purpose of this analysis (another assignment) will be to reveal the main drivers of fuel economy - your analyses should bear this in mind.

Your description of the data should include (but not be restricted to)

• The shape of the distribution of each variable.
• Discussion of the main numerical data summaries and the properties of those summaries.
• Identification of unusual and atypical cases.
• If possible explanations for the atypical cases - use common-sense knowledge of car design.
• Exploration of bivariate relationships, not just with a fuel economy measure but between explanatory variables as well - (try out comparison boxplots and the smoothing technology).
• If it makes sense you can ``build new variables'' (using transformations and by combining other variables) to provide more insight.

You can do this assignment using any statistical package that gets the job done, but if you are using Splus then the following commands are relevant:

Splus commands.

summary, qqnorm, boxplot, hist, plot, spline

Typing in help(function_name) will describe the commands usage.

Graphics windows are opened by the motif() command and closed using the dev.off() command.

in order to print a graph in the Splus UNIX environment proceed as follows:

postscript(``myfile.ps'') # Step 1: specify output file.
plot(x,y) # Step 2: your graphics commands.
dev.off() # step 3: stop writing to the output file.

After you have exited Splus with q(), from UNIX type lp myfile.ps, and you graph should be printing.

Useful fact. Case-wise deletion of missing data is done with the na.omit command, e.g.

car89nomiss <- na.omit(car89)