#
# Unit 8 - Data frames 
#
# Categorical variables:
# A first acquaintance with factors 
# 


# Loading data 
setwd("c:/R")
employees <- read.csv("employees.csv")
print(employees)

# No surprise. 
print(class(employees$GENDER))
# Surprise indeed!
print(class(employees$FNAME))


# Factor is a data structure 
# whose purpose is to storing 
# one-dimensional collections 
# comprised of limited number of 
# unique values, and representing 
# categorical variables, 
# e.g. gender. 
print(employees$GENDER)

# LNAME is not a categorical variable, 
# why?
print(employees$LNAME)
print(class(employees$LNAME))


# Converting a vector into a factor, 
# One common way: the function factor()
# Premise: the conversion is made when 
# the vector represents a categorical variable. 
intGender <- employees$GENDER
print(intGender)
facGender <- factor(employees$GENDER)
print(class(facGender))
print(facGender)

# Setting factor's labels
facGender <- factor(employees$GENDER, 
                    levels = c(1, 2), 
                    labels = c("Female", "Male"))
print(facGender)

# A small demonstration for one 
# scenario in which using factors
# is preferable to using vectors. 
# Bar plot based on employees$gender
intGender <- employees$GENDER
barplot(height = table(intGender), col = c(4, 2))

# Based on employees$gender, factorized 
barplot(height = table(facGender), col = c(4, 2))


# A glimpse onto the reasons 
# to prefer factors over 
# integer vectors. 
# (1) Clearer presentations of data 
# (2) Essential in some statistical procedures, 
# (3) More efficient in terms of memory use.
# (4) Some functions return factors. 
# (5) Using factors it is possible to 
#     display character vectors in a 
#     non-alphabetical order 

# Further observations on factors: 

# Fetching factor's levels: 
# the function levels() 
facGenderLevels <- levels(facGender) 
print(facGenderLevels)
print(class(facGenderLevels))

# You can convert a factor into 
# a vector, e.g. with the 
# function vector() 
# 1
facGender <- factor(employees$GENDER)
print(facGender)
intGender <- as.integer(facGender)
print(intGender)
print(class(intGender))

facGender <- factor(employees$GENDER, 
                    levels = c(1, 2), 
                    labels = c("Female", "Male"))
print(facGender)
intGender <- as.integer(facGender)
print(intGender)


# Note that you can "slice" a factor. 
# The result is itself a factor. 
facGender <- factor(employees$GENDER)
partOfFacGender <- facGender[2:5]
print(partOfFacGender)
print(class(partOfFacGender))

facGender <- factor(employees$GENDER)
print(facGender)
partOfFacGender <- facGender[1:2]
print(partOfFacGender)

# You cannot change factor' contents. 
facGender <- factor(employees$GENDER)
facGender[2] <- 3
print(facGender)

# Factors are not meant to be 
# used in calculations: 
print(facGender + 5)

# and you cannot use factors 
# in comparison expressions: 
if (facGender[2] > 5) {
  print("something")
}

# See also: 
# - gl() - for producing regular 
#          series factors
# - ordered factors 

# So how come R now does not change 
# character columns to factors: 
https://cran.r-project.org/doc/manuals/r-release/NEWS.html