#https://drive.google.com/file/d/1hzSs8JpHb051H2h90swFpEJ7eJ_oILdy/view?usp=sharing

#Working directory
getwd()

5 + 2
5^2

a <- 5

a

a + 2

rm(a)

b <- c(2,2,3,4,5)
c <- c(4,4,6,8,10)

b - c

mean(b)
sd(b)
log.b <- log(b)

mydata <- data.frame(b, c, log.b)

mydata

mydata$b

mean(mydata$b)
median(mydata$b)
sort(table(mydata$b))

hist(mydata$b)

plot(mydata$b ~ mydata$c)
plot(b ~ c, data = mydata)


install.packages("ggplot2")
library(ggplot2)

data(diamonds)

write.csv2(diamonds, "diamonds.csv")
head(diamonds)
str(diamonds)

#find mode, mean and median of price and plot histogram



hist(___)
plot(density(___))

#perform log transformation of price and plot the density
log.price <-  log(diamonds$price)
diamond
plot(___)


###############################
install.packages("dplyr")
install.packages("sjPlot")

install.packages("tidyverse")

library(dplyr)
library(sjPlot)

diamonds %>%
  select(price)

# IMD2003      >>>         EDUCATION
# IMD3010      >>>         SATISFACTION WITH DEMOCRACY

#0. NONE (NO EDUCATION)/ILLITERATE
#1. PRIMARY EDUCATION/LOWER SECONDARY EDUCATION
#2. HIGHER SECONDARY EDUCATION
#3. POST-SECONDARY (NON-UNIVERSITY) EDUCATION 
#4. UNIVERSITY EDUCATION
#6. OTHER [SEE Standalone CSES MODULE CODEBOOK]
#7. VOLUNTEERED: REFUSED
#8. VOLUNTEERED: DON'T KNOW
# 9. MISSING

# 1. VERY SATISFIED
# 2. FAIRLY SATISFIED
# 4. NOT VERY SATISFIED
# 5. NOT AT ALL SATISFIED
# 6. NEITHER SATISFIED NOR DISSATISFIED
# 7. VOLUNTEERED: REFUSED
# 8. VOLUNTEERED: DON'T KNOW
# 9. MISSING

#download data https://drive.google.com/file/d/1qhjPh64IJKIuDdj3W1UIlaBweNRoG18j/view?usp=sharing
#codebook https://cses.org/wp-content/uploads/2020/12/cses_imd_codebook_part2_variables.txt

load("/Volumes/GoogleDrive/Můj disk/Disertace/Data/Kvantitativni/CSES/cses_imd.rdata")

table(cses_imd$IMD2003, cses_imd$IMD3010)

cses_imd <- cses_imd %>%
  mutate(edu = IMD2003,
         dem = IMD3010)

cses_imd_filter <- cses_imd %>%
  filter(edu < 6) %>%
  filter(dem < 6)

cses_imd_filter %>%
  select(edu, dem)

cses_imd_filter %>%
  summarise(avg_edu = mean(edu),
            avg_dem = mean(dem))

cses_imd_filter %>%
  group_by(IMD1004) %>%
  summarise(avg_edu = mean(edu),
            avg_dem = mean(dem))

cses_cz96 <- cses_imd_filter %>%
  filter(IMD1004 == "CZE_1996") 
  
table(cses_imd_filter$edu, cses_imd_filter$dem)

table(cses_cz96$edu, cses_cz96$dem)

chisq.test(table(cses_cz96$edu, cses_cz96$dem))

sjt.xtab(cses_imd_filter$dem, cses_imd_filter$edu)

sjt.xtab(cses_imd_filter$dem, cses_imd_filter$edu, show.exp = T)

sjt.xtab(cses_imd_filter$dem, cses_imd_filter$edu, show.exp = T, show.row.prc = T)

boxplot(cses_cz96$edu ~ cses_cz96$dem)

sjt.xtab(cses_cz96$dem, cses_cz96$edu)

sjt.xtab(cses_cz96$dem, cses_cz96$edu, show.exp = T, show.row.prc = T)

#perform chi-square test for CZ 2010 and variables satisfaction with democracy and gender