Using tRakt

First step: Loading packages and trakt.tv credentials

suppressPackageStartupMessages(library(tRakt)) 
suppressPackageStartupMessages(library(dplyr)) # For convenience
library(ggplot2) # For plotting (duh)
library(knitr)   # for knitr::kable, used to render simple tables

# If you don't have a client.id defined in a key.json, use mine
if (is.null(getOption("trakt.client.id"))){
  get_trakt_credentials(client.id = "12fc1de7671c7f2fb4a8ac08ba7c9f45b447f4d5bad5e11e3490823d629afdf2")
}

Let’s pull some data

# Search via text query
show1  <- trakt.search("Game of Thrones")

# Search via ID (trakt id is used by default)
show2 <- trakt.search.byid(1390) # trakt id of Game of Thrones

# The returned data is identical
identical(show1, show2)

## [1] TRUE

# Search a show and receive basic info
show          <- trakt.search("Breaking Bad")
# Save the slug of the show, that's needed for other functions as an ID
slug          <- show$ids$slug
slug

## [1] "breaking-bad"

# Get the season & episode data
show.seasons  <- trakt.getSeasons(slug) # How many seasons are there?
show.episodes <- trakt.getEpisodeData(slug, show.seasons$season, extended = "full")

# Glimpse at data (only some columns each)
rownames(show.seasons) <- NULL # This shouldn't be necessary
show.seasons[c(1, 3, 4)] %>% kable

show.episodes[c(1:3, 6, 7, 17)] %>% head(10) %>% kable

Some example graphs

Plotting the data is pretty straight forward since I try to return regular data.frames without unnecessary ambiguitiy.

show.episodes$episode_abs <- 1:nrow(show.episodes) # I should probably do that for you.
show.episodes %>%
  ggplot(aes(x = episode_abs, y = rating, colour = season)) +
    geom_point(size = 3.5, colour = "black") +
    geom_point(size = 3) + 
    geom_smooth(method = lm, se = F) +
    labs(title = "Trakt.tv Ratings of Breaking Bad", 
         y = "Rating", x = "Episode (absolute)", colour = "Season")

show.episodes %>%
  ggplot(aes(x = episode_abs, y = votes, colour = season)) +
    geom_point(size = 3.5, colour = "black") +
    geom_point(size = 3) + 
    labs(title = "Trakt.tv User Votes of Breaking Bad Episodes", 
         y = "Votes", x = "Episode (absolute)", colour = "Season")

show.episodes %>%
  ggplot(aes(x = episode_abs, y = scale(rating), fill = season)) +
    geom_bar(stat = "identity", colour = "black", position = "dodge") +
    labs(title = "Trakt.tv User Ratings of Breaking Bad Episodes\n(Scaled using mean and standard deviation)", 
         y = "z-Rating", x = "Episode (absolute)", fill = "Season")

Now some user-specific data

User-specific functions (trakt.user.*) default to user = getOption("trakt.username"), which should have been set by get_trakt_credentials(), so you get your own data per default.
However, you can specifiy any publicly available user. Note that OAuth2 is not supported, so by “publicly available user”, I really mean only non-private users.

# Get a detailed list of shows/episodes I watched
myeps    <- trakt.user.watched(user = "jemus42", type = "shows.extended")

# Get a feel for the data
myeps[c(1:4, 6:7)] %>% 
  arrange(desc(lastwatched.posix)) %>% 
  head(5) %>% 
  kable

# …and the movies in my trakt.tv collection
mymovies <- trakt.user.collection(user = "jemus42", type = "movies")

mymovies %>%
  select(title, year, collected.posix) %>%
  arrange(collected.posix) %>%
  head(5) %>%
  kable

I tried my best to make the returned data as flat and usable as possible.
I tried.

So, well, let’s see: Take watched shows, diff the oldest and youngest lastwatched values to get something like a “watch duration” going and aggregate using it:

myeps %>% 
  group_by(title) %>% 
  summarize(days = round(max(lastwatched.posix) - min(lastwatched.posix))) %>%
  arrange(desc(days)) %>%
  head(10) %>%
  kable

It’s data like this that makes me wish I had been using trakt.tv forever. The potential for interesting data is great, but the limit is, as usual, the source of the data.