the_simpsons_episodes_analysis.R

library(ggplot2)
library(dplyr)

# Basic Analysis
data1 <- read_csv("path")

##Group season and episode
data1_sorted <- data1 %>%
  arrange(season, number_in_season) %>%
  mutate(season_episode = paste(season, number_in_season, sep = ", ")) 


##Repeated values
data1_sorted %>% 
  select(season_episode) %>% 
  group_by(season_episode) %>% 
  tally() %>%
  filter(n > 1)

# IMDB Rating analysis
rating <- data1_sorted %>%
  select(imdb_rating) %>%
  group_by(imdb_rating) %>%
  tally()

## missing values
data1_sorted %>%
  filter(is.na(imdb_rating)) %>%
  tally()

## average rating
data1_sorted %>%
  summarise(imdb_rating_mean = mean(imdb_rating, na.rm = TRUE))

## plot IMDB rating per episode
episode_rating_plot <- ggplot(data1_sorted, aes(season_episode, imdb_rating, group=1)) + 
  geom_line(size=1) + 
  ggtitle("Episodes rating on IMDB") +
  labs(y="IMDB rating", x='season, episode')

## average IMDB rating per season
season_average <- data1_sorted %>%
  group_by(season) %>%
  summarise(imdb_rating_average = mean(imdb_rating, na.rm = TRUE))

## plot IMDB rating per season
season_rating_plot <- ggplot(season_average, aes(Group.1, x)) +
  geom_line(size=1) +
  ggtitle("Average season rating on IMDB") +
  labs(y='IMDB rating', x='season')