reisijad=read_csv("http://www.tlu.ee/~jaagup/andmed/muu/estonia_reisijad.txt")
nrow(reisijad)
reisijad %>% nrow()
reisijad %>% filter(Country=='Estonia') %>% nrow()
nrow(reisijad[reisijad$Country=='Estonia',])/nrow(reisijad)
sum(reisijad$Country=='Estonia')/nrow(reisijad)
round(100*sum(reisijad$Country=='Estonia')/nrow(reisijad))
reisijad %>% filter(Country=='Estonia') %>% 
  {round(100*nrow(.)/nrow(reisijad))}
eestireisijad=filter(reisijad, Country=="Estonia")
eestireisijad %>% filter(Survived==1) %>% nrow()/nrow(eestireisijad)
reisijad %>% group_by(Country) %>% summarise(kogus=n(), k2=mean(Age)) %>%
  arrange(desc(kogus)) %>% filter(kogus>1)

reisijad %>% group_by(Country) %>% 
   summarise(kogus=n(), k2=mean(Age), med=median(Age)) %>%
  arrange(desc(kogus)) %>% filter(kogus>1.5*nrow(reisijad)/100)
reisijad %>% mutate(Ag=floor(Age/10))
reisijad %>% mutate(dAge=Age-mean(Age))
reisijad %>% group_by(Country) %>% mutate(dCAge=Age-mean(Age))
reisijad %>% mutate(Ag=floor(Age/10)) %>% 
             group_by(Ag) %>% summarise(kogus=n())

reisijad %>% filter(Country %in% c("Estonia", "Sweden")) %>% 
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag, Country) %>% summarise(kogus=n())

reisijad %>% filter(Country %in% c("Estonia", "Sweden", "Latvia")) %>% 
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag, Country) %>% summarise(kogus=n()) %>%
  spread(Country, kogus,fill = 0)

reisijad %>% filter(Survived=="1") %>% 
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag) %>% summarise(kogus=n())

reisijad %>% 
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag) %>% summarise(kogus=n(), elus=sum(Survived), p=elus/kogus)

reisijad %>% 
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag) %>% summarise(kogus=n(), naisi=sum(Sex=="F"), p=naisi/kogus)

reisijad %>% filter(Sex=="F", Age<10) %>% nrow()

reisijad %>% filter(Sex=="F") %>%
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag) %>% summarise(kogus=n(), elus=sum(Survived), p=elus/kogus)


reisijad %>% filter(Sex=="M") %>%
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag) %>% summarise(kogus=n(), elus=sum(Survived), p=elus/kogus)

reisijad %>% 
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag, Sex) %>% 
  summarise(kogus=n(), elus=sum(Survived), p=elus/kogus) %>%
  select(Ag, Sex, p) %>%
  spread(Sex, p)


reisijad %>% filter(Country %in% c("Sweden", "Estonia", "Finland")) %>%
  mutate(Ag=floor(Age/10)) %>% 
  group_by(Ag, Country) %>% 
  summarise(kogus=n(), elus=sum(Survived), p=elus/kogus) %>%
  select(Ag, Country, p) %>%
  spread(Country, p)