IMDb TV Show Rating/Votes Plot

imdb_tvshow_ratings_plot.R

install.packages(c("tidyverse", "rvest", "fANCOVA")) 
library(tidyverse)
library(rvest)
library(fANCOVA)
imdb_tvshow_ratings_plot <- function(id, votes=FALSE){
  obj <- paste("http://www.imdb.com/title/", id, "/epdate", sep = "") %>%
    read_html()
  obj <- obj %>%
    html_table(dec = ",", trim = T) %>%
    .[[1]] %>% .[, -5] %>%
    mutate(Title = Episode) %>% select(-Episode) %>%
    separate(col = `#`, into = c("Season", "Episode"))  %>%
    transmute(
      Season = Season %>% factor(levels = order(.)),
      Episode = Episode %>% parse_number,
      `UserRating`,
      Title
    ) %>%
    bind_cols(
      obj %>% html_table(dec = ".", trim = T) %>%
        .[[1]] %>% .[, -5] %>% transmute(UserVotes = UserVotes %>% parse_number())
    ) %>%
    na.omit %>%
    arrange(Season, Episode) %>%
    rownames_to_column() %>%
    mutate(`Episode Number` = rowname) %>%
    select(-rowname) %>%
    mutate(
      UserRating = UserRating %>% parse_number(),
      `Episode Number` = `Episode Number` %>% parse_number()
    )
  if (votes == FALSE) {
    span1 <-
      fANCOVA::loess.as(
        x = obj$`Episode Number`,
        y = obj$UserRating,
        criterion = "gcv"
      )$pars$span
    span2 <-
      fANCOVA::loess.as(
        x = obj$`Episode Number`,
        y = obj$UserRating,
        criterion = "aicc"
      )$pars$span
    span <- exp(mean(log(c(span1, span2))))
  } else{
    span1 <-
      fANCOVA::loess.as(
        x = obj$`Episode Number`,
        y = obj$UserVotes,
        criterion = "gcv"
      )$pars$span
    span2 <-
      fANCOVA::loess.as(
        x = obj$`Episode Number`,
        y = obj$UserVotes,
        criterion = "aicc"
      )$pars$span
    span <- exp(mean(log(c(span1, span2))))
  }
  if (votes == FALSE) {
    obj <-
      obj %>% ggplot(aes(
        x = `Episode Number`,
        y = UserRating,
        color = Season,
        group = Season
      )) +
      scale_y_continuous(limits = c(1, 10), breaks = 1:10) +
      stat_smooth(geom = "line", se = F, alpha = 0.8) +
      geom_point(alpha = 0.5) +
      labs(y = "IMDB Rating") +
      stat_smooth(geom = "line",
                  aes(group = c(0)),
                  alpha = 0.4,
                  span = span)
    obj
  }
  else{
    obj <-
      obj %>% ggplot(aes(
        x = `Episode Number`,
        y = UserVotes,
        color = Season,
        group = Season
      )) +
      scale_y_continuous(limits = c(0, NA)) +
      stat_smooth(geom = "line", se = F, alpha = 0.8) +
      geom_point(alpha = 0.5) +
      labs(y = "Number of votes per episode") +
      stat_smooth(geom = "line",
                  aes(group = c(0)),
                  alpha = 0.4,
                  span = span)
    obj
  }
}

## Example
imdb_tvshow_ratings_plot(id = "tt1266020", 
                         votes = F)+
  ggtitle("Parks and Recreation")