pbiecek

library(survey)
library(ggplot2)
library(dplyr)
library(xtable)
library(tidyr)

pbiecek

# jeżeli nie jest zainstalowany pakiet to instalujemy
# library(devtools)
# install_github('pbiecek/PISA2009lite')
library(PISA2009lite)

studenci <- student2009
szkoly <- school2009

pbiecek

tab <- as.data.frame(table(szkoly$CNT))
names(tab) <- c('Kraje','N')
summary(tab$N)
tab$Kolor <- as.factor(ifelse(tab$Kraje=='Poland',1,2))
ggplot(data = tab,
       aes(x = reorder(Kraje,N),
           y = N,
           fill = Kolor)) +
  geom_bar(stat='identity',
           colour='black') +

pbiecek

tab <- as.data.frame(table(studenci$CNT))
names(tab) <- c('Kraje','N')
summary(tab$N)
tab$Kolor <- as.factor(ifelse(tab$Kraje=='Poland',1,2))
ggplot(data = tab,
       aes(x = reorder(Kraje,N),
           y = N,
           fill = Kolor)) +
  geom_bar(stat='identity',
           colour='black') +

pbiecek

cnames <- names(studenci)
attr(studenci$PV1MATH,'value.labels') <- NULL
attr(studenci$PV2MATH,'value.labels') <- NULL
attr(studenci$PV3MATH,'value.labels') <- NULL
attr(studenci$PV4MATH,'value.labels') <- NULL
attr(studenci$PV5MATH,'value.labels') <- NULL
attr(studenci$W_FSTUWT,'value.labels') <- NULL
for (i in which(cnames=='W_FSTR1'):which(cnames=='W_FSTR80')) {
  attr(studenci[,i],'value.labels') <- NULL
}

pbiecek

badanie <- studenci %>%
  select(CNT, ## kraj
         SCHOOLID, ## identyfikator szkoły
         ST04Q01, ## płeć
         PV1MATH:PV5MATH, ## wartości PV
         W_FSTR1:W_FSTR80, ## wagi replikacyjne
         W_FSTUWT ## wagi ostateczne
         ) %>%
  group_by(CNT) %>% 
  do(schemat = svrepdesign(ids= ~SCHOOLID,

pbiecek

wynik <- badanie %>%
    do(data.frame(svymean(~PV1MATH,design = .$schemat)))
wynik$CNT <- badanie$CNT
names(wynik)[1] <- 'PV1MATH'

pbiecek

### przedział ufności
alpha <- 0.05
z_alpha <- qnorm(1-alpha/2)
wynik$Kolor <- ifelse(wynik$CNT=='Poland','red','black')
### wykresy
ggplot(data = wynik,
       aes(x=reorder(CNT,PV1MATH),
           y=PV1MATH)) +
  geom_errorbar(aes(ymin=PV1MATH-z_alpha*SE,
                    ymax=PV1MATH+z_alpha*SE)) +

pbiecek

wynik_matematyka <- badanie %>%
          do(data.frame(PV1 = svymean(~PV1MATH,design = .$schemat),
                        PV2 = svymean(~PV2MATH,design = .$schemat),
                        PV3 = svymean(~PV3MATH,design = .$schemat),
                        PV4 = svymean(~PV4MATH,design = .$schemat),
                        PV5 = svymean(~PV5MATH,design = .$schemat))) %>%
          ungroup() %>%
          mutate(CNT = badanie$CNT) %>%
          gather(Stat,Value,-CNT) %>%
          separate(Stat,c('PV','STAT'),sep='\\.') %>%

pbiecek

wynik_matematyka$Kolor <- ifelse(wynik_matematyka$CNT=='Poland','red','black')
OECD_mean <- mean(wynik_matematyka$PV)
### wykresy
ggplot(data = wynik_matematyka,
       aes(x=reorder(CNT,PV),
           y=PV)) +
  geom_errorbar(aes(ymin=PV-z_alpha*SE,
                    ymax=PV+z_alpha*SE)) +
  geom_point(colour='red') +
  theme_bw() +