#5710 Lab Week of 2/23 R Script
#The following reads it in from the web. If you do something in Rcmdr you will have to select charity as the active dataset.

charity <- read.table("http://www.unt.edu/rss/class/mike/data/charity.txt", header=TRUE, sep="", na.strings="NA", dec=".", strip.white=TRUE)


#1. Conduct multiple regression
RegModel.1 <- lm(GIVEN~CASH+GENDER+IMPORT, data=charity)
summary(RegModel.1)

#2. Obtain confidence intervals for raw coefficients.
confint(RegModel.1, level=.95)

#3. Obtain graphs of the unique relationships of each predictor with GIVEN.
av.plots(RegModel.1, ask=FALSE, identify.points=TRUE)

#4. Calculate relative importance.
library(relaimpo)
calc.relimp(RegModel.1, rela=T)

#5. a) Graphical diagnostic tests. Shown here just for giggles, Rcmdr menu is much more efficient.
oldpar <- par(oma=c(0,0,3,0), mfrow=c(2,2))
plot(RegModel.1)
par(oldpar)

#5. b) Statistical test of normality of residuals.
shapiro.test(RegModel.1$res)

#6. b) Statistical test of homoscedasticity. Requires lmtest package if you aren't doing it in Rcmdr menus.
bptest(GIVEN ~ CASH + GENDER + IMPORT, varformula = ~ fitted.values(RegModel.1), studentize=FALSE, data=charity)

#7. a) Graphical test of linearity.
cr.plots(RegModel.1, ask=FALSE)

#b) Statistical test of linearity. Requires lmtest package if you aren't doing it in Rcmdr menus.
resettest(GIVEN ~ CASH + GENDER + IMPORT, power=2:3, type="regressor", data=charity)

#8. Variance inflation factors to test assumption of collinearity.
vif(RegModel.1)

#9. Conduct robust regression.
library(MASS)
summary(rlm(GIVEN ~ CASH + GENDER + IMPORT, data=charity))

#10. Validate the model and obtain bias-adjusted R square.
library(Design)
myvalidatedmodel=ols(GIVEN ~ CASH + GENDER + IMPORT, x=T, y=T, data=charity)
validate(myvalidatedmodel)