EDA_titanic.R

library(tidyverse)
library(randomForest)
library(ggthemes)

######################################################
# Reading in the data
######################################################

train <- read.csv("train.csv", stringsAsFactors = F)
test <- read.csv("test.csv", stringsAsFactors = F)

glimpse(train)
summary(train)

sum(is.na(train))

# Bind together the two datasets
full <- bind_rows(train %>% mutate(dset = "train"),
                  test %>% mutate(dset = "test")
                  )
full <- full %>% mutate(dset = factor(dset))

str(full)

######################################################
# Feature engineering
######################################################

full$Title <- gsub('(.*, )|(\\..*)', '', full$Name)

# Show title counts by sex
table(full$Sex, full$Title)

# Titles with very low cell counts to be combined to "rare" level
rare_title <- c('Dona', 'Lady', 'the Countess','Capt', 'Col', 'Don', 
                'Dr', 'Major', 'Rev', 'Sir', 'Jonkheer')


# Also reassign mlle, ms, and mme accordingly
full$Title[full$Title == 'Mlle']        <- 'Miss' 
full$Title[full$Title == 'Ms']          <- 'Miss'
full$Title[full$Title == 'Mme']         <- 'Mrs' 
full$Title[full$Title %in% rare_title]  <- 'Rare Title'

table(full$Sex, full$Title)


# Finally, grab surname from passenger name
full$Surname <- sapply(full$Name, function(x) strsplit(x, split = '[,.]')[[1]][1])

# Check how many distinct surnames we have
nlevels(factor(full$Surname))

# Create a family size variable including the passenger themselves
full$Fsize <- full$SibSp + full$Parch + 1

table(full$Fsize)

# Create a family variable 
full$Family <- paste(full$Surname, full$Fsize, sep='_')

# Family size vs. survival
ggplot(full[(full$dset=="train"),], aes(x = Fsize, fill = factor(Survived))) +
  geom_bar(stat='count', position='dodge') +
  scale_x_continuous(breaks=c(1:11)) +
  labs(x = 'Family Size', y = 'Count') +
  theme_few()

head(full$Cabin, 30)

#TODO
strsplit(full$Cabin[2],NULL)

######################################################
# Missing values imputation
######################################################


######################################################
# Modelling
######################################################