#CHAPTER 4: THE R SOFTWARE IN ACTION
######################################################

####################################################
####################################################
#4.2: “R” INSTALLATION AND SESSOION 
####################################################

#INSTALL PACKAGES
#install.packages("ggm")
#install.packages("polycor")
#install.packages("Hmisc")
#install.packages("psych")
#install.packages("ggplot2")
#install.packages("rlm")
#install.packages("MASS")
#install.packages("pastecs")

#LIBRARIES TO LOAD
library(ggm); library(Hmisc); library(polycor); library(psych)
library(ggplot2); library(rlm); library(MASS); library(pastecs)

####################################################
####################################################
#4.3: BASICS OF R DEMONSTRATED
####################################################
#4.3.2: IMPORT DATA TO THE SESSION
####################################################
#READ CSV FILE INTO SESSSION
#<Path> provided by the user to inform of location of data 
TaskData <- read.csv("C:\\<PATH>\\ScoreComplexPrep.csv", 
	header=TRUE, sep=",")

####################################################
#4.3.3: KNOW THY DATA ANALYSIS
####################################################
#INSPECT FIRST SIX CASES
head(TaskData)
#INSPECT MAKEUP OF VARIABLES
str(TaskData)

#DESCRIPTIVE STATISTICS
summary(TaskData)
#ADDITIONAL DESCRIPTIVE STATISTICS 
describe(TaskData)

#PAIRS PANEL
pairs.panels(TaskData[c("Score","Complex","Prep")])

#Q-Q GRAPHIC FOR NORMALITY
par(mfrow = c(2, 2))
anx<- qqnorm(TaskData$Complex, main = "Q-Q Complexity "); qqline(TaskData$Complex)
rev<- qqnorm(TaskData$Prep, main = "Q-Q Preparation "); qqline(TaskData$Prep)
exm<- qqnorm(TaskData$Score, main = "Q-Q Efficiency Score "); qqline(TaskData$Score)
par(mfrow = c(1, 1))

####################################################
#4.3.4: CORRELATION ANALYSIS
####################################################
#CREATE DATAFRAME FROM THE THREE VARIABLES OF INTEREST
TaskData2<- TaskData[,c("Score","Complex","Prep")]
head(TaskData2)

#CORRELATION WITH PEARSON AND SPEARMAN
cor(TaskData2, method="pearson")
cor(TaskData2, method="spearman")

#CONVERT TASKDATA2 TO A MATRIX
TaskMatrix<- as.matrix(TaskData2)

#DETERMINE P-VALUES TO CORRELATIONS
rcorr(TaskMatrix, type = "pearson")
rcorr(TaskMatrix, type = "spearman")

#INTERVAL TO CORRELATION WITH PEARSON
cor.test(TaskData$Complex, TaskData$Score, method="pearson")
cor.test(TaskData$Complex, TaskData$Prep, method="pearson")
cor.test(TaskData$Prep, TaskData$Score, method="pearson")

####################################################
#4.3.5: PARTIAL CORRELATION ANALYSIS
####################################################
#R SQUARED
cor(TaskData2, method="pearson")^2
cor(TaskData2, method="spearman")^2

#PARTIAL CORRELATION TO SCORE AND COMPLEX
cor(TaskData2, method="pearson")
pcExAn<- pcor(c("Score", "Complex", "Prep"), var(TaskData2))  
pcExAn
pcExAn^2
pcor.test(pcExAn, 1, 103)

#ALTERNATIVE PARTIAL SCORE TO COMPLEX
pcExRe<- pcor(c("Exam", "Revise", "Anxiety"), var(examData2)) 
pcExRe
pcExRe^2
pcor.test(pcExRe, 1, 103)

####################################################
#4.3.6: REGRESSION ANALYSIS
####################################################
##Conduct linear model to test for interaction between predictor variables

#MODEL OF MAIN EFFECTS
TaskPred <- lm(Score ~ Complex + Prep, TaskData)
summary(TaskPred)
confint(TaskPred)

#MODEL WITH INTERACTION
TaskPredIntr <- lm(Score ~ Complex * Prep, TaskData)  
summary(TaskPredIntr)
confint(TaskPredIntr)

#ROBUST MODEL FOR NONNORMAL DATA
rlm.Tasklm <- rlm(Score ~ Complex + Prep, data = TaskData)
summary(rlm.Tasklm)