#set working directory
setwd("D:/AJG/Documents/AJG Documents/R/R/RData")
getwd()

#1

#install and load packages
install.packages("psych")
install.packages()
library('psych')

#import data
jobsat <- read.csv("job-satisfaction.csv", header=TRUE, colClasses=c("numeric", "numeric", "numeric", "numeric", "numeric", "numeric", "numeric"))
jobsat <- read.csv("job-satisfaction.csv", header=TRUE, colClasses=c("numeric"))
describe(jobsat)
boxplot(jobsat)

#test data for suitability for factor analysis
cor(jobsat)
jobsat.cor <- cor(jobsat)
jobsat.cor
round(jobsat.cor ,3)

KMO(jobsat)
cortest.bartlett(jobsat)
det(jobsat.cor)

#calculate eigenvalues
jobsat.eigen <- eigen(jobsat.cor)
jobsat.eigen$values

#draw a scree plot
scree(jobsat, fa=FALSE)
scree(jobsat, main="Scree Plot")

#carry our parallel analysis
fa.parallel(jobsat, fa="pc", show.legend=TRUE, main="Scree plot with parallel analysis")

#extract factors
jobsat.analysis <- principal(jobsat, nfactor=3, rotate="none")
jobsat.analysis

#with rotation
jobsat.analysis <- principal(jobsat, nfactor=3, rotate="varimax")

#show diagrams
fa.diagram(jobsat.analysis)
fa.diagram(jobsat.analysis, simple=FALSE)
fa.diagram(jobsat.analysis, digits=3)
fa.diagram(jobsat.analysis, cut=.7)
fa.diagram(jobsat.analysis, simple=FALSE, l.cex=3)
fa.diagram(jobsat.analysis, simple=FALSE, adj=4)

#correlation matrix from model
factor.model(jobsat.analysis$loadings)
factor.residuals(jobsat.cor, jobsat.analysis$loadings)
jobsat.residuals <- factor.residuals(jobsat.cor, jobsat.analysis$loadings)

#draw histogram,
hist(jobsat.residuals)

#to get component scores
jobsat.analysis <- principal(jobsat, nfactor=3, rotate="varimax", scores=TRUE)
jobsat.anaysis$scores
jobsat.scores <- jobsat.analysis$scores
jobsat.scores.round <- round(jobsat.scores,2)

#add to original data frame
jobsatplus <- cbind(jobsat,jobsat.scores.round)

#change column names
colnames(jobsatplus)[c(8,9,10)] <- c("Component 1", "Component 2", "Component 3")

#to use just some variables
jobsat1 = jobsat[, c(2:9)] or c(2,4,6)

#2

#import data and look at it
studsatall <- read.csv("student-satisfaction.csv", header=TRUE, colClasses=c("numeric"))
studsatall <- read.csv("student-satisfaction-text.csv", header=TRUE)
studsat <- studsatall[, c(6:17)] 
describe(studsat)
boxplot(studsat)

#look for high correlations
studsat.cor <- cor(studsat)
studsat.cor

#determine eigenvalues
studsat.eigen <- eigen(studsat.cor)
studsat.eigen$values

#various tests
KMO(studsat)
cortest.bartlett(studsat)
det(studsat.cor)

#how many factors
scree(studsat, fa=FALSE)
scree(studsat, main="Scree Plot")
fa.parallel(studsat, fa="pc", show.legend=TRUE, main="Scree plot with parallel analysis")

#exctract factors
studsat.analysis <- principal(studsat, nfactor=2, rotate="none")
studsat.analysis <- principal(studsat, nfactor=2, rotate="varimax")

#diagram
fa.diagram(studsat.analysis)
fa.diagram(studsat.analysis, simple=FALSE)
fa.diagram(studsat.analysis, digits=3)
fa.diagram(studsat.analysis, cut=.7)
fa.diagram(studsat.analysis, simple=FALSE, l.cex=3)
fa.diagram(studsat.analysis, simple=FALSE, adj=4)

#correlations from model compared to original
factor.model(studsat.analysis$loadings)
factor.residuals(studsat.cor, studsat.analysis$loadings)

#draw histogram
studsat.residuals <- factor.residuals(studsat.cor, studsat.analysis$loadings)
hist(studsat.residuals)

#analyse again with scores=TRUE
studsat.analysis <- principal(studsat, nfactor=2, rotate="varimax", scores=TRUE)
studsat.scores <- studsat.analysis$scores

#add to original dataframe
studsatplus <-cbind(studsat,studsat.scores)

#change column names
colnames(studsatplus)[14] <- "Component 2"
colnames(studsatplus)[c(13,14)] <- c("Component 1", "Component 2")

#t test
t.test(studsatplus$RC1 ~ studsatplus$gender)

#anova
anova1 <- aov(studsatplus$ethnicgp ~ studsatplus$RC1)
summary(anova1)

---------------

#change values in dataframe
df[df == 'Old Value'] <- 'New value'

studsatall$gender[studsatall$gender == "1"] <- "Male"
studsatall$gender[studsatall$gender == "2"] <- "Female"

#check classes
as.data.frame(sapply(studsatall,class))

#change integer to numeric
studsatall[,6:17] <- lapply(studsatall[,6:17], as.numeric)