Titanic Assignment

```{r }

library(readr) T3 <- read_csv(“https://goo.gl/At238b”) View(T3)

titanic <- data.frame(T3\(survived, T3\)sex,T3\(age, T3\)sibsp, T3\(parch, T3\)fare,T3$embarked) titanic View(titanic)

feature_name<- c(‘survived’, ‘sex’, ‘age’, ‘sibsp’, ‘parch’, ‘embarked’, ‘fare’) colnames(titanic)<- feature_name View(titanic)

#Remove NAs if any

any(is.na(titanic)) titanic_na<-na.omit(titanic) any(is.na(titanic_na))

#Perform a statistical analysis of the titanic dataset.

summary(titanic) sapply(titanic_na, sd) summary(titanic_na\(survived) summary(titanic_na\)age) summary(titanic_na\(sibsp) summary(titanic_na\)fare) summary(titanic_na$parch)

```{r }

#Survived is the dependent variable, find its proportion in the dataset.
table(titanic_na$survived)
prop.table(table(titanic_na$survived))
str(titanic_na$survived)

#Make Survived, embarked, and sex as factors

titanic$survived <- as.factor(titanic$survived)

titanic$embarked <- as.factor(titanic$embarked)

titanic$sex <- as.factor(titanic$sex)


#Find the correlation matrix between survival and the other features.

cor(titanic_na$survived, titanic_na[, c("embarked", "sibsp", "parch")])



#Plot survival with other features to see if any correlation exists.

plot(titanic$survived~titanic$sex)

```{r } #Set a seed to 1000 and use it to split the dataset into 80% training, 20% testing.

set.seed(1000)

train_index <- sample(1:nrow(titanic_na), 0.8 * nrow(titanic_na))

train <- titanic_na[train_index, ]

test<- titanic_na[-train_index, ]

#Build your training (till index 1046) and test (till index 1308)datasets

titanic_train <- titanic_na[1:834,] titanic_test <- titanic_na[835:1043,]

#Download the rpart for “Recursive Partitioning and RegressionTrees” if it is not in your R base and bring it in yourenvironment. library(rpart)

#Train your learner on the training dataset and save your model in a variable fitfit <- rpart(Survived ~ Sex + Age + SibSp + Parch + Fare + Embarked,data=train, method=“class”)

fit <- rpart(survived ~ sex + age + sibsp + parch + fare + embarked, data = titanic_train, method = “class”)

fit

```{r }

install.packages(c("rattle", "rpart.plot", "RColorBrewer"))

library(rpart.plot)
library(rattle)
library(RColorBrewer)

fancyRpartPlot(fit)

#Examine the tree you obtained. What is the most “important”feature over which the tree first split? Do you agree with theTitanic slogan “ Women and children first!”


#The most important feature which the tree first split would be the "sex" feature. I do agree with with the slogan "Women and children first!"

#Use the predict function with your model fit to makepredictions on the test dataset and save it in a variablePrediction:Prediction <- predict(fit, test, type = "class")

Prediction <- predict(fit, titanic_test, type = "class")
Prediction


#Save the results in a data frame Results that will have twocolumns 1: PassengerSex =test$Sex, and 2: Survived = Prediction

Results <- data.frame(PassengerSex = titanic_test$sex, Survived = Prediction)
Results

#Save your data frame in a .csv file by usingwrite.csv(Results, file = "Titanicdtree.csv", row.names = FALSE)

write.csv(Results, file = "Titanicdtree.csv", row.names = FALSE)