*
* Example BIGCITY.DAT for pp 95-109
*
open data bigcity.dat
data(format=free,org=columns) 1 12 sales ads compete
*
scatter(footer="Figure 3-2 Big City Bookstore",hlabel="Advertising ($)",vlabel="Sales ($)")
# ads sales
*
linreg sales
# constant ads
*
scatter(footer="Figure 3-3 Big City Bookstore",hlabel="Advertising ($)",vlabel="Sales ($)",line=%beta)
# ads sales
*
prj fitted
set pcterr = 100.0*%resids/sales
*
print(picture="*.###") / sales fitted %resids pcterr
*
* @reganova is a regression "post-processor" which does the ANOVA table based
* upon the last regression.
*
@reganova
*
* page 118-119.
* PRJ can compute quite a few other statistics, in addition to just the fitted
* values. In particular, the STDERRS option returns the individual forecast
* standard errors, while XVX gives the value under the square root in formula
* (3-22). This needs to multiplied by %seesq (the regression variance), and
* square rooted in order to get the standard error of the regression line.
*
prj(stderrs=stderrs,xvx=sysq)
set fiddle = sqrt(%seesq*sysq)
print / fitted stderrs fiddle