# Dateien im Projekt herunterladen
download.file(url = "https://rstats.kamapu.net/Ressourcen/Dateien/KursDateien.zip",
              destfile = "KursDateien.zip")
unzip("KursDateien.zip", overwrite = TRUE)
unlink("KursDateien.zip")

# Daten Umstrukturieren ----
Bonn <- readRDS("BonnBevoelkerung.rds")
Bezirke <- readRDS("BonnBezirke.rds")

Bonn <- merge(Bonn, Bezirke)

Bonn2021 <-subset(Bonn, Jahr == 2021)
Bonn2021 <- Bonn2021[
  c("BezirkNr", "Frauen", "Maenner")]
Bonn2021$FrauenProzent <- with(Bonn2021, {
  Frauen/(Maenner + Frauen)*100
})

# Das Gleiche als Workflow
library(dplyr)

Bonn2019 <- Bonn %>%
  filter(Jahr == 2019) %>%
  select(BezirkNr, Frauen, Maenner) %>%
  mutate(FrauenProzent =
           Frauen/(Frauen + Maenner)*100)

library(tidyr)

Gender <- Bonn %>%
  mutate(Frauen = Frauen/Gesamt*100,
         Maenner = Maenner/Gesamt*100) %>%
  pivot_longer(cols = c("Maenner", "Frauen"),
               names_to = "Gender",
               values_to = "Prozent") %>%
  select(BezirkNr, Jahr, Gender, Prozent)

library(ggplot2)

ggplot(data = Gender, aes(x = Gender, y = Prozent,
                          fill = Gender)) +
  geom_boxplot() +
  facet_wrap(~Jahr)

# Alles in Workflow
Bonn %>%
  mutate(Frauen = Frauen/Gesamt*100,
         Maenner = Maenner/Gesamt*100) %>%
  pivot_longer(cols = c("Maenner", "Frauen"),
               names_to = "Gender",
               values_to = "Prozent") %>%
  select(BezirkNr, Jahr, Gender, Prozent) %>%
  ggplot(aes(x = factor(Jahr), y = Prozent,
             fill = factor(Jahr))) +
  geom_boxplot() +
  facet_wrap(~Gender)

# Regressionen -----
?airquality
data("airquality")

summary(airquality)

air <- subset(airquality,
          !is.na(Ozone) & !is.na(Solar.R))
pairs(air)

plot(Ozone ~ Wind, data = air, pch = 16)

model_1 <- lm(Ozone ~ Wind, data = air)

summary(model_1)

# Model im Graphiken ----
plot(Ozone ~ Wind, data = air, pch = 16)
abline(model_1, lty = "dashed", col = "red")

# Was ist im Objekt?
str(model_1)
str(summary(model_1))

r_quadrat <- summary(model_1)$r.squared
r_quadrat
text(15, 150, labels = round(r_quadrat, 2), col = "blue")

# Mit Funktionen
coef(model_1)
residuals(model_1)
fitted.values(model_1)

# Predicted vs. Observed
plot(fitted.values(model_1) ~ Ozone,
     data = air, pch = 16,
     xlim = c(0, 200), ylim = c(0, 200))
abline(a = 0, b = 1, lty = "dashed", col = "red")

# Residuen
plot(resid(model_1) ~ Ozone, data = air, pch = 16)
abline(h = 0, lty = "dashed", col = "red")

# QQplots
plot(model_1, which = 2)

hist(air$Ozone)
with(air, hist((Ozone - mean(Ozone))/sd(Ozone)))

plot(model_1, which = 2)

qqnorm(resid(model_1))
qqline(resid(model_1), lty = "dashed", col = "red")

# Multidimensionales Model ----
model_2 <- lm(Ozone ~ Wind + Temp, data = air)
summary(model_2)