library(psych)
library(tidyverse)
library(GGally)
library(qgraph)

# Načtení dat
df <- haven::read_sav("https://is.muni.cz/go/j6bdrc")
df <- df %>% 
  select(id:neshody)

glimpse(df)

# Extracting metadata

metadata <- tibble(
  var_names = names(df[-1]),
  var_labels = map_chr(df[-1], 
                       ~attributes(.)$label)
)

metadata %>% 
  print(n = Inf)

# Korelace mezi optimismem a životní spokojeností - grafy
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_point()

df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_point() +
  theme_classic(base_size = 15)

# geom_point() a position_jitter
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_point(position = position_jitter(width = .4, height = .1,
                                        seed = 123))

# geom_jitter()
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = .1, width = .3)


# theme_classic(base_size = 15)
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = .1, width = .3) +
  theme_classic(base_size = 15)

theme_set(theme_classic(base_size = 15))

df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = .1, width = .3)

# Adding a regression line
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = 0.1, width = .3) +
  geom_smooth()

# Customizing the regression line
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = 0.1, width = .3) +
  geom_smooth(method = "lm",
              level = .99,
              color = "red",
              fill = "red",
              alpha = .2)

# Adding labels
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = 0.1, width = .3) +
  geom_smooth(method = "lm",
              level = .99,
              color = "red",
              fill = "red",
              alpha = .2) +
  labs(x = "Optimismus",
       y = "Životní spokojenost")

# Customizing scales
df %>% 
  ggplot(aes(optim, ziv_sp)) +
  geom_jitter(height = 0.1, width = .3) +
  geom_smooth(method = "lm",
              level = .99,
              color = "red",
              fill = "red",
              alpha = .2) +
  labs(x = "Optimismus",
       y = "Životní spokojenost") +
  scale_x_continuous(breaks = seq(5, 25, by = 2)) +
  scale_y_continuous(breaks = seq(1, 4, by = .5)) 

# Base R functions
cor.test(df$optim, df$ziv_sp)
cor.test(df$optim, df$ziv_sp,
         method = "spearman")
cor.test(df$optim, df$ziv_sp,
         conf.level = .99)

variables <- c("ziv_sp", "optim", "selfe", "effi", "zdravi", "deprese")
df_cor <- select(df, all_of(variables))

cor(df_cor)
cor(df_cor,
    use = "pairwise")
cor(df_cor,
    use = "pairwise") %>% 
  round(digits = 2)

# GGally a funkce ggpairs()
ggpairs(df, 
        columns = variables) 

plot_scatter <- function(data, mapping){
  ggplot(data = data, mapping = mapping) + 
    geom_jitter(alpha = .05) + 
    geom_smooth(color = "blue",
                se = FALSE,
                method = "lm",
                linetype = "11") +
    geom_smooth(color = "red",
                fill = "red",
                alpha = .2)
}

ggpairs(df, 
        columns = variables,
        lower = list(continuous = plot_scatter)) 

plot_hist <- function(data, mapping){
  ggplot(data = data, mapping = mapping) + 
    geom_histogram(fill = "grey",
                   color = "black")
}

ggpairs(df, 
        columns = variables,
        lower = list(continuous = plot_scatter),
        diag = list(continuous = plot_hist)) 


# Q-Q plots and histograms
df_long <- df %>% 
  select(all_of(variables)) %>% 
  pivot_longer(everything(), 
               names_to = "variable") %>% 
  mutate(value = as.double(scale(value)),
         .by = variable)

df_long %>% 
  ggplot(aes(value)) +
  geom_histogram(aes(y = after_stat(density)),
                 binwidth = .25) +
  facet_wrap(~variable) +
  stat_function(fun = dnorm,
                color = "red", 
                linewidth = 1)

df_long %>% 
  ggplot(aes(sample = value)) +
  geom_qq_line() +
  geom_qq() +
  facet_wrap(~variable)

# Funce corr.test() z balíčku psych
out <- psych::corr.test(df_cor)
print(out, short = FALSE)
str(out)
out$stars


# Use complete cases only
out <- psych::corr.test(df_cor,
                        use = "complete")
print(out, short = FALSE)

psych::corr.test(df_cor)
psych::corr.test(df_cor, 
                 adjust = "holm", # jakou korekci p-hodnot použít
                 method = "spearman", # typ korelace
                 alpha = 0.005) %>% # hladina spolehlivosti bude 1 - alpha
  print(short = FALSE)

str(out)
out$stars
print(out, short = FALSE)

# Let's say you have 3 comparisons and want 95% overall confidence (α = 0.05):

# First interval: 1 - (0.05/3) = 0.983 → 98.3% CI
# Second interval: 1 - (0.05/2) = 0.975 → 97.5% CI
# Third interval: 1 - (0.05/1) = 0.95 → 95% CI


# Korelogramy pomocí balíčku corrplot a stejnojmenné funkce corplot()
r_mat <- cor(df_cor, use = "pairwise")
r_mat # korelační matice

corrplot::corrplot(r_mat)
corrplot::corrplot(r_mat, 
                   method = "number", 
                   order = "hclust")
corrplot::corrplot.mixed(r_mat, 
                         lower = "number",
                         upper = "ellipse",
                         lower.col = "black",
                         order = "hclust")

# Network plots
df_cor <- df %>% 
  select(warm_m:neshody)

names(df_cor)

r_mat <- cor(df_cor,
             use = "pairwise",
             method = "spearman")

diag(r_mat) <- 0

qgraph(r_mat,
       layout = "spring",
       theme = "colorblind",
       threshold = .1,
       repulsion = .7,
       groups = list(VŘE = 1:2,
                     DŮV = 4:5,
                     IND = 7:9,
                     NEG = 10:12,
                     CRO = 14:16,
                     CSK = 17:20,
                     VM = 26:28),
       labels = c("VŘ.M", "VŘ.O", "MON", "DUV.R", "DUV.V", "VZsR", "IND.1", "IND.2", 
                  "IND.3", "NEG.1", "NEG.2", "NEG.3", "ROD.E", "ROD.P", "ROD.A", "R.S",
                  "ŠK.E", "ŠK.P", "ŠK.A", "ŠK.S", "OPT", "ZSP", "EST", "EFF",
                  "PSS", "VM.U", "VM.S", "VM.N", "DEP", "NESH"),
       GLratio = 9)