#  Measuring the influence of one variable on another: Regression

mydataframe <- read.delim("http://www.robin-beaumont.co.uk/virtualclassroom/book2data/regression1.dat", header=TRUE)
names(mydataframe)
mydataframe

hba1c_model <- lm(HBA1C~fastingBG, data=mydataframe)
summary(hba1c_model)
Confint(hba1c_model, level=0.95)

#standard scatterplot with line of best fit
library(car) 
scatterplot(fastingBG~HBA1C, reg.line=lm, smooth=FALSE, spread=TRUE, 
  boxplots=FALSE, span=0.5, data=mydataframe)
 
# For CI and prediction intervals on the graph
# use the UsingR package
 
  install.packages("UsingR", dependencies=TRUE)
  library(UsingR)
 # show.ci  adds both CI and prediction lines
 simple.lm(mydataframe$fastingBG, mydataframe$HBA1C, show.ci=TRUE)
 
 #  R squared CI
   install.packages("psychometric", dependencies=TRUE)
  library(psychometric)
   CI.Rsqlm(hba1c_model )

### Influence measures
influence.measures(hba1c_model)

### Residuals
resid(hba1c_model)
rstandard(hba1c_model)  
rstudent(hba1c_model)
### plotting residuals
plot(resid(hba1c_model)) 
plot(rstandard(hba1c_model)) 

## rownames for subsetting
rownames(mydataframe) 

hba1c_model2 <- update(hba1c_model1, subset=(rownames(mydataframe) != 12))
summary(hba1c_model2)