R

hw2Mac.r

# -------------------------------
# Homework 2 - Machine Learning 
# Samet Sait Talayhan
# -------------------------------
#      o                    / \  //\
#       o    |\___/|      /   \//  \\
#            /0  0  \__  /    //  | \ \    
#           /     /  \/_/    //   |  \  \  
#           @_^_@'/   \/_   //    |   \   \ 
#           //_^_/     \/_ //     |    \    \
#        ( //) |        \///      |     \     \
#      ( / /) _|_ /   )  //       |      \     _\
#    ( // /) '/,_ _ _/  ( ; -.    |    _ _\.-~        .-~~~^-.
#  (( / / )) ,-{        _      `-.|.-~-.           .~         `.
# (( // / ))  '/\      /                 ~-. _ .-~      .-~^-.  \
# (( /// ))      `.   {            }                   /      \  \
#  (( / ))     .----~-.\        \-'                 .~         \  `. \^-.
#             ///.----..>        \             _ -~             `.  ^-`  ^-_
#               ///-._ _ _ _ _ _ _}^ - - - - ~                     ~-- ,.-~
#                                                                  /.-~
#
# Note: I created this beauty dragon using cowsay app, ~$ cowsay -f dragon "Talayan"

## Step 2: Exploring and preparing the data ---- 

#import the csv file
data_set <- read.csv("hw2train.csv", stringsAsFactors = FALSE)

# examine the structure of the test data frame
str(data_set)

# drop the class features,
#data_set <- data_set[,16:43]

# shuffle data
data_set <- data_set[sample(nrow(data_set)),]

# table of rightmost feature
table(data_set$Feat.43)

# recode Feat.43 as a factor
data_set$Feat.43 <- factor(data_set$Feat.43, levels = c(-1, 1),
                         labels = c("Negative", "Positive"))

# table or proportions with more informative labels
round(prop.table(table(data_set$Feat.43)) * 100, digits = 1)

# summarize three numeric features
summary(data_set[c("Feat.17", "Feat.18", "Feat.19")])

# create normalization function
normalize <- function(x) {
  return ((x - min(x)) / (max(x) - min(x)))
}

# test normalization function - result should be identical
normalize(c(1, 2, 3, 4, 5))
normalize(c(10, 20, 30, 40, 50))

# normalize the homework data
data_set_n <- as.data.frame(lapply(data_set[1:42], normalize))

# confirm that normalization worked
summary(data_set_n$Feat.17)

# shuffle normalized data
#data_set_n_shuffle <- data_set_n[sample(nrow(data_set_n)),]

# create training and test data
data_set_train <- data_set_n[1:99, ]
data_set_test <- data_set_n[100:164, ]


# table of
table(data_set$Feat.43)

# create labels for training and test data
data_set_train_labels <- data_set[1:99, 43]
data_set_test_labels <- data_set[100:164, 43]

## Step 3: Training a model on the data ----

# load the "class" library
library(class)

data_set_test_pred <- knn(train = data_set_train, test = data_set_test,
                      cl = data_set_train_labels, k=10)


## Step 4: Evaluating model performance ----

# load the "gmodels" library
library(gmodels)

# Create the cross tabulation of predicted vs. actual
CrossTable(x = data_set_test_labels, y = data_set_test_pred,
           prop.chisq=FALSE)


## Step 5: Optimize the solution

# load the "corrplot" library, 
# the library to compute correlation matrix
install.packages("corrplot")
library(corrplot)
library(mlbench)
library(caret)

# compute the correlation matrix
correlation_mat <- cor(data_set_n)

# summurize the correlation matrix
print(correlation_mat)

# find attributes that are highly corrected (ideally >0.75)
highlyCorrelated <- findCorrelation(correlationMatrix, cutoff=0.5)

# visualize the matrix, clustering features by correlation index.
corrplot(correlation_mat, order = "hclust")

setwd("/home/t4/Dropbox/Machine Learning/hw2")
install.packages("plyr_1.8.2", repos=NULL)