Spotify

Let’s analyze the ‘spotify’ data:

spotify_all <- read_csv('https://bcdanl.github.io/data/spotify_all.csv')

The data.frame spotify_all includes information about Spotify users’ playlists.

• The unit of observation in spotify_all is a track in a music playlist.

Variable Description

• pid: playlist ID; unique ID for playlist

• playlist_name: a name of playlist

• pos: a position of the track within a playlist (starting from 0)

• artist_name: name of the track’s primary artist

• track_name: name of the track

• duration_ms: duration of the track in milliseconds

• album_name: name of the track’s album

Q1a

• Find the ten most popular song.
  • A value of a song is defined as a combination of a artist_name value and a track_name value.
  • Who are artists for those ten most popular song?

Answer:

popular_songs <- spotify_all |> 
  group_by(artist_name, track_name) |> 
  summarize(Count = n()) |> 
  arrange(desc(Count))

top_ten_songs <- popular_songs |> 
  head(n = 10)

top_ten_songs_artists <- top_ten_songs$artist_name

Q1b

• Find the five most popular artist in terms of the number of occurrences in the data.frame, spotify_all.
• What is the most popular song for each of the five most popular artist?

Answer:

popular_artists <- spotify_all |> 
  group_by(artist_name) |> 
  summarize(Count = n()) |> 
  arrange(desc(Count))

top_five_artists <- popular_artists |> 
  head(n = 5)

top_song_by_artist <- popular_songs |> 
  filter(artist_name %in% top_five_artists$artist_name) |> 
  group_by(artist_name) |> 
  slice_head(n = 1)

Q1c

• Create a data.frame named drake_one_dance that includes all playlists featuring the song where track_name equals “One Dance” and artist_name equals “Drake”.
  • Ensure that the data.frame drake_one_dance does not contain any playlists where the song “One Dance” by Drake is absent.
• What is the song with the highest frequency of appearances after Drake’s “One Dance” in the newly created data.frame, drake_one_dance?

Answer:

drake_one_dance_playlists <- spotify_all |> 
  filter(track_name == "One Dance" & artist_name == "Drake") |> 
  select(pid)

drake_one_dance <- spotify_all |> 
  filter(pid %in% drake_one_dance_playlists$pid)

check_drake_one_dance <- drake_one_dance %>%
  group_by(pid) %>%
  filter(all(c("One Dance" %in% track_name, "Drake" %in% artist_name))) %>%
  ungroup()

popular_song_after_one_dance <- drake_one_dance |> 
  group_by(track_name) |> 
  summarize(Appearances = n()) |> 
  arrange(desc(Appearances)) |> 
  filter(track_name != "One Dance") |> 
  slice_head(n = 1)

Q1d

Provide both (1) ggplot code and (2) a couple of sentences to describe the relationship between pos and the ten most popular artists.

Answer:

top_ten_artist_names <- popular_artists |> 
  head(n = 10)

top_ten_artists <- spotify_all |> 
  filter(artist_name %in% top_ten_artist_names$artist_name)

top_ten_artists <- left_join(top_ten_artists, top_ten_artist_names, by = "artist_name")

ggplot(top_ten_artists,
       aes(x = pos))+
  geom_bar(position = "stack") +
  facet_wrap(artist_name ~ .)

# The top ten artists all have a similar relationship with pos. They all have a higher frequency of when pos is at 0 and close to 0 and decrease in frequency as pos increases. Since these are the top ten artists, it makes sense that their songs would be at the beginning of playlists more often than at the end. 

Q1e

Create the data.frame with pid-artist level of observations with the following four variables:

• pid: playlist id
• playlist_name: name of playlist
• artist: name of the track’s primary artist, which appears only once within a playlist
• n_artist: number of occurrences of artist within a playlist

Answer:

playlist_artist_count <- spotify_all |> 
  group_by(pid, artist_name) |> 
  summarize(n_artist = n()) |> 
  ungroup()

playlist_artist_level <- spotify_all %>%
  select(pid, playlist_name, artist_name) %>%
  distinct() %>%
  left_join(playlist_artist_count, by = c("pid", "artist_name"))

playlist_artist_level <- rename(playlist_artist_level, artist = artist_name)