train <- read.csv("../input/train.csv",stringsAsFactor=FALSE)
test  <- read.csv("../input/test.csv",stringsAsFactor=FALSE)

train$Sex[which(train$Sex=="female")]<-1
train$Sex[which(train$Sex=="male")]<-0

for(i in c("Master.","Miss.","Mrs.","Mr.","Dr.")){
train$Name[grep(i,train$Name,fixed=TRUE)]<-i}

masterage<-mean(train$Age[which(train$Name=="Master.")],trim=.5,na.rm=TRUE)
missage<-mean(train$Age[which(train$Name=="Miss.")],trim=.5,na.rm=TRUE)
mrsage<-mean(train$Age[which(train$Name=="Mrs.")],trim=.5,na.rm=TRUE)
mrage<-mean(train$Age[which(train$Name=="Mr.")],trim=.5,na.rm=TRUE)
drage<-mean(train$Age[which(train$Name=="Dr.")],trim=.5,na.rm=TRUE)

train$Age[is.na(train$Age)&train$Name=="Master."]<-masterage
train$Age[is.na(train$Age)&train$Name=="Miss."]<-missage
train$Age[is.na(train$Age)&train$Name=="Mrs."]<-mrsage
train$Age[is.na(train$Age)&train$Name=="Mr."]<-mrage
train$Age[is.na(train$Age)&train$Name=="Dr."]<-drage

train$Child<- 0
train[which(train$Age<14),c("Child")]<-1

train$Family<- NA

for(i in 1:nrow(train)){
train$Family[i]<-train$SibSp[i]+train$Parch[i]+1
}

train$Mother<- 0
train[which(train$Parch>0 & train$Age>18 & train$Name=="Mrs."),c("Mother")]<-1

train$Cabin[which(!train$Cabin=="")]<-1
train$Cabin[which(train$Cabin=="")]<-0

test$Sex[which(test$Sex=="female")]<-1
test$Sex[which(test$Sex=="male")]<-0

for(i in c("Master.","Miss.","Mrs.","Mr.","Dr.")){
test$Name[grep(i,test$Name,fixed=TRUE)]<-i}

masterage<-mean(test$Age[which(test$Name=="Master.")],trim=.5,na.rm=TRUE)
missage<-mean(test$Age[which(test$Name=="Miss.")],trim=.5,na.rm=TRUE)
mrsage<-mean(test$Age[which(test$Name=="Mrs.")],trim=.5,na.rm=TRUE)
mrage<-mean(test$Age[which(test$Name=="Mr.")],trim=.5,na.rm=TRUE)
drage<-mean(test$Age[which(test$Name=="Dr.")],trim=.5,na.rm=TRUE)

test$Age[is.na(test$Age)&test$Name=="Master."]<-masterage
test$Age[is.na(test$Age)&test$Name=="Miss."]<-missage
test$Age[is.na(test$Age)&test$Name=="Mrs."]<-mrsage
test$Age[is.na(test$Age)&test$Name=="Mr."]<-mrage
test$Age[is.na(test$Age)&test$Name=="Dr."]<-drage
test$Age[which(is.na(test$Age))]<- 20

test$Child<- 0
test[which(test$Age<14),c("Child")]<-1

test$Family<- NA

for(i in 1:nrow(test)){
test$Family[i]<-test$SibSp[i]+test$Parch[i]+1
}

test$Mother<- 0
test[which(test$Parch>0 & test$Age>18 & test$Name=="Mrs."),c("Mother")]<-1

test$Cabin[which(!test$Cabin=="")]<-1
test$Cabin[which(test$Cabin=="")]<-0

train.glm<- glm(Survived~Age+Child+Family+Sex*Pclass+Cabin,family=binomial,data=train)
p.hats<- predict.glm(train.glm,newdata=test,type="response")

survival <- NA
for(i in 1:length(p.hats)) {
  if(p.hats[i] > .5) {
    survival[i] <- 1
  } else {
    survival[i] <- 0
  }
}

kaggle.sub <- cbind(test$PassengerId,survival)
colnames(kaggle.sub) <- c("PassengerId", "Survived")
write.csv(kaggle.sub, file = "kaggle.csv", row.names = FALSE)