Homework 14. Due Monday, Nov. 2.
================================

Hand in edited R code and (small) portions of results 
doing in R the following calculations and graphics 
accomplished in SAS in parts (a)-(d) of the class-handout 
which you can find in the web-page handout BassSAS.txt 
within the directory

http://www.math.umd.edu/~evs/s798c/Slogs/BassSAS.txt

(i) Input the data from 

http://www.math.umd.edu/~evs/s430.old/Data/bass

entering it into a data-frame with columns

ID Lake Alk pH  Calc Chlor AvMrc nsamp ,

fitting simple linear regression for response  logAvM = log(AvMrc)
in terms of predictor  Alk  ,  and overplotting on the scatterplot
(with different line-types and/or colors)

the predicted line  and

the predicted line plus or minus 1.96 times the residual SE 
(see RlogF09.LinRegr.txt or a regression text for the formula.)

and "identify" (on the plot) the points lying outside the confidence
band you drew in this way.

(ii) Remove the most extreme outliers (your choice), and re-fit 
the model using forward stepwise selection with an entry 
criterion alpha = .05, using the variables Alk pH Calc Chlor

Present an ANOVA table for your final fitted model, and a 
graphical display of some kind showing how it fits better 
than the simple linear regression  logAvM  versus  Alk.