## Title: "Titanic Survival Prediction via RF"
#### Author: Rahima Karimova (Inspired by David Langer's Titanic 
#### Data Analysis video https://www.youtube.com/watch?v=32o0DnuRjfg 
#### and kaggler-Megan Risdal)
#### Date: 29.11.2016

### Necessary packages 
#### library(tidyverse) 
#### library(ggthemes)
#### library(stringr)
#### library(scales)
#### library(mice)
#### library(ranger)
### library(ranger)
```{r include=FALSE}
library(tidyverse) #Cleaning, visualization...
library(ggthemes) # visualisation
library(stringr)
library(scales)
library.path <- cat(.libPaths())
library("mice", lib.loc = library.path)
library(mice) # NA replacement
library(ranger)
library(caret)
library(randomForest) # forecast
```
# Reading your data
```{r}
train <- read.csv('../input/train.csv', stringsAsFactors = F)
test  <- read.csv('../input/test.csv', stringsAsFactors = F)
```


## Preparing train and test for binding
```{r}
test.survived<-data.frame(Survived= rep("None", nrow(test)), test[,])
head(test.survived)
ntest.survived<-test.survived[,c(2,1,3,4,5,6,7,8,9,10,11,12)]
head(ntest.survived)
names(train)
names(ntest.survived)
```

## Binding data sets
```{r}
data.combined<-rbind(train,ntest.survived)
glimpse(data.combined)

data.combined$Survived<-as.factor(data.combined$Survived)
data.combined$Pclass<-as.factor(data.combined$Pclass)
glimpse(data.combined)
```
## Visualisation part 1


# Feature extraction

### Any valuable information in the Title (Mrs, Miss, Mr, Master)?

```{r}
misses<-data.combined[which(str_detect(data.combined$Name,"Miss.")),]
head(misses)
mrses<-data.combined[which(str_detect(data.combined$Name,"Mrs.")),]
mrses[1:5,]
masters<-data.combined[which(str_detect(data.combined$Name,"Master")),]
masters[1:5,]
mr<-data.combined[which(str_detect(data.combined$Name,"Mr.")),]
mr[1:5,]
```
## Adding new variable - Title
```{r}
extractTitle<-function(Name){
  Name<-as.character(Name)
  if(length(grep("Miss.",Name))>0){
    return("Miss.")
  } else if (length(grep("Master.",Name))>0){
    return("Master.")
  } else if(length(grep("Mrs.",Name))>0){
    return("Mrs.")
  } else if(length(grep("Mr.", Name))>0){
    return("Mr.")
  } else {
    return("Other")
  }
}
titles<-NULL
for (i in 1:nrow(data.combined)){
  titles<-c(titles,extractTitle(data.combined[i,"Name"]))
}
```
### Ok. Lets factorise our new feature
```{r}
data.combined$Title<-as.factor(titles)
```

### Factoring Sibsp variable to further continue with histogram plotting
```{r}
data.combined$SibSp<-as.factor(data.combined$SibSp)
```
## Visualization of the survival rate by Sibsp, Title and Pclass
```{r,echo=T}
g2.5<-ggplot(data.combined[1:891,],aes(x=SibSp, fill=Survived))+geom_bar(width=0.5)+
facet_wrap(~Pclass+Title)+
  ggtitle("Pclass,Title")+
  xlab("Sibsp")+
  ylab("Total Count")+
  ylim(0,300)+
  labs(fill="Survived")

g2.5
```

### Factoring Parch variable
```{r}
data.combined$Parch<-as.factor(data.combined$Parch)
```
## Visualization of the Survival rate by Parch, Title and Pclass
``` {r,echo=T}
g2.6<-ggplot(data.combined[1:891,],aes(x=Parch, fill=Survived))+geom_bar(width=1)+
   facet_wrap(~Pclass+Title)+
   ggtitle("Pclass,Title")+
   xlab("Parch")+
   ylab("Total Count")+
   ylim(0,300)+
   labs(fill="Survived")
  
g2.6
``` 
## Creating the Family Size feature
```{r}
  temp.sibsp<-c(train$SibSp,test$SibSp)
  temp.parch<-c(train$Parch,test$Parch)
  data.combined$Family.Size<-as.factor(temp.sibsp+temp.parch+1)
```
```{r,echo=T}
g2.7<-ggplot(data.combined[1:891,],aes(x=Family.Size, fill=Survived))+geom_bar(width=1)+
    facet_wrap(~Pclass+Title)+
    ggtitle("Pclass,Title")+
    xlab("Family.Size")+
    ylab("Total Count")+
    ylim(0,300)+
    labs(fill="Survived")
  
  g2.7
```
# Working on NAs.
##Constructing subgroups
```{r}
subsmr<-subset(data.combined[1:891,], is.na(Age) & Title=="Mr.")
summary(subsmr)
```
```{r,echo=T}
g3.1<-ggplot(subsmr,aes(x=Pclass,fill=Survived))+geom_bar(width=0.5)+facet_wrap(~Family.Size)
g3.1
```   
```{r}  

```
### Here we see the age range is wide so we need futher classification
  
  
## Substituting NAs with means
```{r} 
   data.combined2<-data.combined
   
   data.combined2$Embarked[c(62,830)]
   data.combined2$Embarked[c(62,830)]<-"C"
   data.combined2[1044,]
```   
### Replacement of missing fare value with median fare for class/embarkment
```{r}
   data.combined2$Fare[1044]<-median(data.combined2[data.combined2$Pclass==3 & data.combined2$Embarked=="S",]$Fare,na.rm=T)
   #Make variables factors into factors
   factor_vars <- c('PassengerId','Pclass','Sex','Embarked',
                    'Title','Family.Size')
   
   data.combined2[factor_vars] <- lapply(data.combined2[factor_vars], function(x) as.factor(x))
```
```{r}
   ### Setting a random seed
   set.seed(129)
```
### Performing mice imputation, excluding certain less-than-useful variables:
```{r}
mice_mod <- mice(data.combined2[, !names(data.combined2) %in% c('PassengerId','Name','Ticket','Cabin','Survived')], method='rf') 
### complete output saving
```{r}
   mice_output <- complete(mice_mod)
```  
   ## Ploting age distributions
```{r}
par(mfrow=c(1,2))
   hist(data.combined2$Age, freq=F, main='Age: Original Data', 
        col='darkgreen', ylim=c(0,0.04))
   hist(mice_output$Age, freq=F, main='Age: MICE Output', 
        col='lightgreen', ylim=c(0,0.04))
```
   
### Replacement of Age variable from the mice model.
```{r}
data.combined2$Age <- mice_output$Age
```

### Replacement of Deck variable from the mice model.
  ```{r}
data.combined2$Deck <- mice_output$Deck
summary(data.combined2)
```
# Prediction

## Splitting the data back into a train set and a test set
```{r}
   train <- data.combined2[1:891,]
   test <- data.combined2[892:1309,]
```   
### Set a random seed
```{r}
   set.seed(754)
```
   
   ## Building the model (note: not all possible variables are used)
   ```{r}
   rf_model <- train(factor(Survived) ~ Pclass + Sex + Age + SibSp + Parch + 
                              Fare + Embarked + Title,data=train,
                              method="ranger",
                              trControl = trainControl(method = "cv", number = 5, verboseIter = TRUE
))
   
   ```
  rf_model
  plot(model)
  ## Model Evaluation 
  max(rf_model[["results"]])
   #Prediction