Hello, and welcome to my very first Markdown publication. The vast majority of my time spent with R has been for university courses, so I thought it would be a good change of pace to apply the skills I’ve learned to a personal project. I have learned a great deal from books like Hadley Wickham’s R for Data Science, as well as tutorials, articles, and markdown pages from sites such as RPubs, R-bloggers, and Towards Data Science. My hope is that this project can in turn be used to help someone else down the road, as others have helped me.
In this project, we will be exploring how to use Spotify’s API and the spotifyr package to access data about a particular artists discography. We’ll also be utilizing the geniusr package to retrieve lyrics for songs, allowing for sentiment to be explored and visualized, and doing our best to apply a clustering algorithm to the discography.
To begin, we will first need to head over to the Spotify for Developers page, where we will be registering an application to obtain an API key. Once you’ve logged in, select “Create an App” and fill out the required fields. Completing this will give you access to two important fields: your client id and your client secret (or API key). These fields will be used to let the API know who is accessing it and that you have proper authentication. To get started, first install the spotifyr package if you don’t already have it.
install.packages("spotifyr")Next, we can use the following code to pass out authentication credentials to the API, giving us access.
Sys.setenv(SPOTIFY_CLIENT_ID = 'your client id')
Sys.setenv(SPOTIFY_CLIENT_SECRET = 'your client secret')
access_token <- get_spotify_access_token()Now that you are authenticated with the API, we can begin using the spotifyr package’s functions to retrieve information. In this project, we will be analyzing Mac Miller’s discography and exploring how his sound changed throughout his career. To get information regarding a single artist’s discography, we can use the get_artist_audio_features() function. The function will return a dataframe containing information about all of the artists’ music that is hosted on Spotify. The function takes many possible arguments, but for the sake of this project, we only need to pass in one: the name of the artist for which we wish to get information.
library(spotifyr)
mm_data <- get_artist_audio_features(artist = "Mac Miller")To check that this function performed as expected, let’s take a very quick glance at the returned dataframe.
colnames(mm_data)## [1] "artist_name" "artist_id"
## [3] "album_id" "album_type"
## [5] "album_images" "album_release_date"
## [7] "album_release_year" "album_release_date_precision"
## [9] "danceability" "energy"
## [11] "key" "loudness"
## [13] "mode" "speechiness"
## [15] "acousticness" "instrumentalness"
## [17] "liveness" "valence"
## [19] "tempo" "track_id"
## [21] "analysis_url" "time_signature"
## [23] "artists" "available_markets"
## [25] "disc_number" "duration_ms"
## [27] "explicit" "track_href"
## [29] "is_local" "track_name"
## [31] "track_preview_url" "track_number"
## [33] "type" "track_uri"
## [35] "external_urls.spotify" "album_name"
## [37] "key_name" "mode_name"
## [39] "key_mode"unique(mm_data$album_name)## [1] "Faces"
## [2] "Circles (Deluxe)"
## [3] "Circles"
## [4] "Swimming"
## [5] "The Divine Feminine"
## [6] "Best Day Ever (5th Anniversary Remastered Edition)"
## [7] "GO:OD AM"
## [8] "Live From Space"
## [9] "Watching Movies with the Sound Off (Deluxe Edition)"
## [10] "Watching Movies with the Sound Off"
## [11] "Mac Miller : Live From London (With The Internet)"
## [12] "Macadelic (Remastered Edition)"
## [13] "Blue Slide Park (Commentary Version)"
## [14] "Blue Slide Park (Edited Version)"
## [15] "Blue Slide Park"
## [16] "K.I.D.S. (Deluxe)"
## [17] "K.I.D.S."head(mm_data$track_name, 15)## [1] "Inside Outside"
## [2] "Here We Go"
## [3] "Friends (feat. ScHoolboy Q)"
## [4] "Angel Dust"
## [5] "Malibu"
## [6] "What Do You Do (feat. Sir Michael Rocks)"
## [7] "It Just Doesn’t Matter"
## [8] "Therapy"
## [9] "Polo Jeans (feat. Earl Sweatshirt)"
## [10] "Happy Birthday"
## [11] "Wedding"
## [12] "Funeral"
## [13] "Diablo"
## [14] "Ave Maria"
## [15] "55"dim(mm_data)## [1] 305 39Excellent! The returned dataframe contains 305 observations, or in this case songs, and each observation has 39 variables.
We can see from the unique(mm_data$album_name) function that the albums are listed in order of upload date, with Faces being the most recent album added to Mac’s Spotify page. While this may seem handy, it is important to note that the order in which albums are uploaded to Spotify is not always the same order that the albums were released. We can observe this by taking a look at the album_release_date variable.
library(tidyverse)
library(plotly)mm_data %>%
select(album_name, album_release_date) %>%
distinct()## # A tibble: 17 x 2
## album_name album_release_date
## <chr> <chr>
## 1 Faces 2021-10-15
## 2 Circles (Deluxe) 2020-03-19
## 3 Circles 2020-01-17
## 4 Swimming 2018-08-03
## 5 The Divine Feminine 2016-09-16
## 6 Best Day Ever (5th Anniversary Remastered Edition) 2016-06-03
## 7 GO:OD AM 2015-09-18
## 8 Live From Space 2013-12-17
## 9 Watching Movies with the Sound Off (Deluxe Edition) 2013-06-18
## 10 Watching Movies with the Sound Off 2013-06-18
## 11 Mac Miller : Live From London (With The Internet) 2013-01-01
## 12 Macadelic (Remastered Edition) 2012-03-23
## 13 Blue Slide Park (Commentary Version) 2011-11-15
## 14 Blue Slide Park (Edited Version) 2011-11-15
## 15 Blue Slide Park 2011-11-08
## 16 K.I.D.S. (Deluxe) 2010-08-13
## 17 K.I.D.S. 2010-08-13This readout implies that Faces is the most recent album to release. However, by checking Mac’s discography, we can see that Faces was actually released as a mixtape back in 2014, much earlier than the variable from Spotify’s data would suggest. Since this project involves analyzing how Mac’s music changed throughout the duration of his career, it is important to have accurate ordering of dates associated with the albums.
To remedy this, we can use some quick web scraping to pull the release dates from the wiki page and amend the data.
# run install.packages('rvest') if you don't have this package already
library(rvest)
url <- "https://en.wikipedia.org/wiki/Mac_Miller_discography"
wp <- read_html(url)
rel_dates <- html_nodes(
wp, "th i , .plainrowheaders th+ td li:nth-child(1) , th i a") %>%
html_text()Now, the above code may look a little intimidating if you are new to webscraping. That’s okay, it looks much scarier than it really is. The read_html() function simply reads the page’s html code and stores it as a list in your R environment. Then, we need to tell R what parts of the webpage we want extracted. To do this, I used the SelectorGadget extension for Chrome. Using the tool makes webscraping very simple, you just highlight the elements you wish to capture and the tool will give you the CSS selector for it. That is how the arguments you see in the html_nodes() function were found. Once you’ve identified the nodes, pass the results to the html_text() function and voila! You know have text from a website stored right in your R environment.
Let’s check out what our scraping resulted in:
head(rel_dates, 20)## [1] "Blue Slide Park" "Blue Slide Park"
## [3] "Released: November 8, 2011[16]" "Watching Movies with the Sound Off"
## [5] "Watching Movies with the Sound Off" "Released: June 18, 2013[20]"
## [7] "GO:OD AM" "GO:OD AM"
## [9] "Released: September 18, 2015[22]" "The Divine Feminine"
## [11] "The Divine Feminine" "Released: September 16, 2016[24]"
## [13] "Swimming" "Swimming"
## [15] "Released: August 3, 2018[26]" "Circles"
## [17] "Circles" "Released: January 17, 2020[31]"
## [19] "Live from Space" "Live from Space"While that looks pretty good, you can see that the album titles are listed twice, and the release dates could be formatted a little nicer. Let’s fix that up to a nicer format.
# Remove duplicates and format into dataframe for manipulation
rel_dates <- matrix(
unique(rel_dates), ncol = 2, byrow = T) %>%
as.data.frame()
# Filter out any works that aren't hosted on Spotify
rel_dates <- rel_dates %>% filter(
V1 %>% tolower() %in% (unique(mm_data$album_name) %>%
gsub("( \\().*", "", .) %>%
tolower()))
# Cleaning date text
rel_dates$V2 <- gsub("(?:Released: )", "", rel_dates$V2)
rel_dates$V2 <- gsub(".{4}$", "", rel_dates$V2)
# Converts textual dates to date type object
rel_dates$V2 <- lubridate::parse_date_time(
rel_dates$V2, orders = "mdy") %>%
lubridate::as_date()
# Check results to ensure they look as expected
rel_dates## # A tibble: 11 x 2
## V1 V2
## <chr> <date>
## 1 Blue Slide Park 2011-11-08
## 2 Watching Movies with the Sound Off 2013-06-18
## 3 GO:OD AM 2015-09-18
## 4 The Divine Feminine 2016-09-16
## 5 Swimming 2018-08-03
## 6 Circles 2020-01-17
## 7 Live from Space 2013-12-17
## 8 K.I.D.S. 2010-08-13
## 9 Best Day Ever 2011-03-11
## 10 Macadelic 2012-03-23
## 11 Faces 2014-05-11Now our release dates are in a more workable format. However, before we merge the two datasets, let’s first take one more look at the album names in our main dataframe, mm_data.
unique(mm_data$album_name)## [1] "Faces"
## [2] "Circles (Deluxe)"
## [3] "Circles"
## [4] "Swimming"
## [5] "The Divine Feminine"
## [6] "Best Day Ever (5th Anniversary Remastered Edition)"
## [7] "GO:OD AM"
## [8] "Live From Space"
## [9] "Watching Movies with the Sound Off (Deluxe Edition)"
## [10] "Watching Movies with the Sound Off"
## [11] "Mac Miller : Live From London (With The Internet)"
## [12] "Macadelic (Remastered Edition)"
## [13] "Blue Slide Park (Commentary Version)"
## [14] "Blue Slide Park (Edited Version)"
## [15] "Blue Slide Park"
## [16] "K.I.D.S. (Deluxe)"
## [17] "K.I.D.S."We see from this readout that many albums contain multiple editions, such as deluxe releases, remasters, or commentary bonuses. To prevent our analysis from being biased towards those repeated works, we should selectively filter out albums that are listed multiple times. Firstly, the commentary version of Blue Slide Park will be omitted. Next, for any album that has a deluxe release, we will keep only the deluxe release, dropping the original album from the dataset. Lastly, we will rename Best Day Ever (5th Anniversary Remastered Edition) and Macadelic (Remastered Edition). This will help us when merging the datasets.
# Shortening 'Best Day Ever' album name for merging
mm_data$album_name[
mm_data$album_name ==
"Best Day Ever (5th Anniversary Remastered Edition)"] = "Best Day Ever"
# Dropping '(Remastered Edition)' from Macadelic
mm_data$album_name[
mm_data$album_name ==
"Macadelic (Remastered Edition)"] = "Macadelic"
# Drop any non-deluxe editions of albums that have deluxe editions
# Note that Blue Slide Park's additional versions were also dropped
# Live From London was dropped as it only included already present songs
mm_data <- filter(mm_data, !(album_name %in% c(
"Circles",
"Watching Movies with the Sound Off",
"K.I.D.S.",
"Mac Miller : Live From London (With The Internet)",
"Blue Slide Park (Commentary Version)",
"Blue Slide Park (Edited Version)")))
# Adding " (Deluxe)" onto album names in true release date set for merging
for(i in c(2, 6, 8)){
if(i == 2){
rel_dates$V1[i] = str_c(rel_dates$V1[i], " (Deluxe Edition)")}
else
rel_dates$V1[i] = str_c(rel_dates$V1[i], " (Deluxe)")}
# Adjusting Capitalization of "Live from Space" to match mm_data$album_name
rel_dates$V1[rel_dates$V1 == "Live from Space"] = "Live From Space"Now that our extra versions have been dropped from the data, we can finally merge our two datasets to attach the accurate release dates.
# Performing the merge of the two datasets
mm_data <- left_join(mm_data, rel_dates,
by = c("album_name" = "V1"))
mm_data <- mm_data %>% rename("true_rel_date" = "V2")With the merge complete, let’s take a look at the differences between the original album_release_date column and our new true_rel_date column.
select(mm_data, album_release_date, album_name, true_rel_date) %>%
unique()## # A tibble: 11 x 3
## album_release_date album_name true_rel_date
## <chr> <chr> <date>
## 1 2021-10-15 Faces 2014-05-11
## 2 2020-03-19 Circles (Deluxe) 2020-01-17
## 3 2018-08-03 Swimming 2018-08-03
## 4 2016-09-16 The Divine Feminine 2016-09-16
## 5 2016-06-03 Best Day Ever 2011-03-11
## 6 2015-09-18 GO:OD AM 2015-09-18
## 7 2013-12-17 Live From Space 2013-12-17
## 8 2013-06-18 Watching Movies with the Sound Off (Deluxe ~ 2013-06-18
## 9 2012-03-23 Macadelic 2012-03-23
## 10 2011-11-08 Blue Slide Park 2011-11-08
## 11 2010-08-13 K.I.D.S. (Deluxe) 2010-08-13Excellent! Now that we have each album’s true release date, we can go ahead and drop the old variable album_release_date and rename our new variable to take its place.
mm_data <- select(mm_data, -album_release_date) %>%
rename(album_release_date = true_rel_date)Before we head into the next step in our analysis, let’s first make sure that we don’t have multiple entries of any tracks.
mm_data$track_name[duplicated(mm_data$track_name)]## [1] "Congratulations (feat. Bilal)"
## [2] "Dang! (feat. Anderson .Paak)"
## [3] "Stay"
## [4] "Skin"
## [5] "Cinderella (feat. Ty Dolla $ign)"
## [6] "Planet God Damn (feat. Njomza)"
## [7] "Soulmate"
## [8] "We (feat. CeeLo Green)"
## [9] "My Favorite Part"
## [10] "God Is Fair, Sexy Nasty (feat. Kendrick Lamar)"
## [11] "Doors"
## [12] "Brand Name"
## [13] "Rush Hour"
## [14] "Two Matches (feat. Ab-Soul)"
## [15] "100 Grandkids"
## [16] "Time Flies (feat. Lil B)"
## [17] "Weekend (feat. Miguel)"
## [18] "Clubhouse"
## [19] "In the Bag"
## [20] "Break the Law"
## [21] "Perfect Circle / God Speed"
## [22] "When in Rome"
## [23] "ROS"
## [24] "Cut the Check (feat. Chief Keef)"
## [25] "Ascension"
## [26] "Jump"
## [27] "The Festival (feat. Little Dragon)"Good thing we checked! 27 duplicate entries is no joke, so let’s figure out how this happened. Familiarity with Spotify and Mac’s library leads me to initially suspect that these tracks appear on both an explicit and clean versions of their respective albums. We can check this assumption rather quickly, so let’s do so.
mm_data %>%
filter(
track_name %in% (mm_data$track_name[duplicated(mm_data$track_name)])) %>%
select(track_name, album_name, explicit) %>%
arrange(album_name)## # A tibble: 54 x 3
## track_name album_name explicit
## <chr> <chr> <lgl>
## 1 Doors GO:OD AM TRUE
## 2 Brand Name GO:OD AM TRUE
## 3 Rush Hour GO:OD AM TRUE
## 4 Two Matches (feat. Ab-Soul) GO:OD AM TRUE
## 5 100 Grandkids GO:OD AM TRUE
## 6 Time Flies (feat. Lil B) GO:OD AM TRUE
## 7 Weekend (feat. Miguel) GO:OD AM TRUE
## 8 Clubhouse GO:OD AM TRUE
## 9 In the Bag GO:OD AM TRUE
## 10 Break the Law GO:OD AM TRUE
## # ... with 44 more rowsAs expected, it looks like the duplicate entries stem from clean editions of albums. To handle this, we can first double-check that aside from the clean albums, every album in the dataset contains at least one explicit song. If so, then we can group the observations by album and simply drop any album that contains no explicit tracks.
# note we use album_id here because it is unique for explicit and clean versions
mm_data %>%
group_by(album_id, album_name) %>%
count(explicit == TRUE) %>%
arrange(album_name)## # A tibble: 19 x 4
## album_id album_name `explicit == T~` n
## <chr> <chr> <lgl> <int>
## 1 13fsGE9UN5VaAkETSs94un Best Day Ever TRUE 16
## 2 6VhDYmsjHqRxKXd0z7hmXI Blue Slide Park TRUE 16
## 3 1YZ3k65Mqw3G8FzYlW1mmp Circles (Deluxe) FALSE 12
## 4 1YZ3k65Mqw3G8FzYlW1mmp Circles (Deluxe) TRUE 2
## 5 5SKnXCvB4fcGSZu32o3LRY Faces FALSE 1
## 6 5SKnXCvB4fcGSZu32o3LRY Faces TRUE 24
## 7 2Tyx5dLhHYkx6zeAdVaTzN GO:OD AM TRUE 17
## 8 6lEUoXk2C9IpUWPd4caiNE GO:OD AM FALSE 17
## 9 5pL6fzBD4sLs9hyau2CeUi K.I.D.S. (Deluxe) FALSE 2
## 10 5pL6fzBD4sLs9hyau2CeUi K.I.D.S. (Deluxe) TRUE 16
## 11 0oPKygNJATeXkPWre0R0Nr Live From Space TRUE 14
## 12 7nVdkG4gZZxB1I1RLN27fJ Macadelic FALSE 2
## 13 7nVdkG4gZZxB1I1RLN27fJ Macadelic TRUE 15
## 14 5wtE5aLX5r7jOosmPhJhhk Swimming FALSE 1
## 15 5wtE5aLX5r7jOosmPhJhhk Swimming TRUE 12
## 16 4gtXD5SL0yysd1eRIrDpnZ The Divine Feminine FALSE 10
## 17 6f6tko6NWoH00cyFOl4VYQ The Divine Feminine FALSE 2
## 18 6f6tko6NWoH00cyFOl4VYQ The Divine Feminine TRUE 8
## 19 3T02fCxAjApu18taJLLbyN Watching Movies with the Sound~ TRUE 19We can see here that the only albums with two different entries for album_id are GO:OD AM and The Divine Feminine. These entries represent the explicit and clean editions of the albums. To handle this, we can simply filter out any entry that matches the album_id value of the clean editions of these albums.
mm_data <- mm_data %>%
filter(!album_id %in% c("6lEUoXk2C9IpUWPd4caiNE", "4gtXD5SL0yysd1eRIrDpnZ"))Now that our release date and duplicate entry issues have been solved, let’s take a look at which variables we’d like to keep, and which ones we can exclude moving forward.
print(names(mm_data))## [1] "artist_name" "artist_id"
## [3] "album_id" "album_type"
## [5] "album_images" "album_release_year"
## [7] "album_release_date_precision" "danceability"
## [9] "energy" "key"
## [11] "loudness" "mode"
## [13] "speechiness" "acousticness"
## [15] "instrumentalness" "liveness"
## [17] "valence" "tempo"
## [19] "track_id" "analysis_url"
## [21] "time_signature" "artists"
## [23] "available_markets" "disc_number"
## [25] "duration_ms" "explicit"
## [27] "track_href" "is_local"
## [29] "track_name" "track_preview_url"
## [31] "track_number" "type"
## [33] "track_uri" "external_urls.spotify"
## [35] "album_name" "key_name"
## [37] "mode_name" "key_mode"
## [39] "album_release_date"As we can see, there are quite a few variables in the set. To get an idea of what each variable holds, we can use tidyverse’s glimpse() function.
glimpse(mm_data)## Rows: 179
## Columns: 39
## $ artist_name <chr> "Mac Miller", "Mac Miller", "Mac Miller",~
## $ artist_id <chr> "4LLpKhyESsyAXpc4laK94U", "4LLpKhyESsyAXp~
## $ album_id <chr> "5SKnXCvB4fcGSZu32o3LRY", "5SKnXCvB4fcGSZ~
## $ album_type <chr> "album", "album", "album", "album", "albu~
## $ album_images <list> [<data.frame[3 x 3]>], [<data.frame[3 x ~
## $ album_release_year <dbl> 2021, 2021, 2021, 2021, 2021, 2021, 2021,~
## $ album_release_date_precision <chr> "day", "day", "day", "day", "day", "day",~
## $ danceability <dbl> 0.689, 0.489, 0.557, 0.628, 0.591, 0.665,~
## $ energy <dbl> 0.7460, 0.8670, 0.6780, 0.8220, 0.5810, 0~
## $ key <int> 0, 6, 1, 5, 6, 10, 8, 1, 1, 11, 1, 0, 1, ~
## $ loudness <dbl> -5.971, -4.668, -5.772, -5.439, -5.760, -~
## $ mode <int> 1, 0, 1, 1, 0, 1, 0, 0, 1, 0, 1, 1, 1, 0,~
## $ speechiness <dbl> 0.0549, 0.4660, 0.2790, 0.2340, 0.1650, 0~
## $ acousticness <dbl> 0.07710, 0.02050, 0.14900, 0.46700, 0.145~
## $ instrumentalness <dbl> 0.00e+00, 0.00e+00, 0.00e+00, 0.00e+00, 4~
## $ liveness <dbl> 0.2910, 0.3530, 0.6110, 0.4860, 0.1120, 0~
## $ valence <dbl> 0.376, 0.550, 0.366, 0.410, 0.226, 0.444,~
## $ tempo <dbl> 113.569, 75.100, 148.103, 78.029, 78.104,~
## $ track_id <chr> "2EFqMCOdTTkcFYHoJH21Jr", "40dlJFdqfm8Cay~
## $ analysis_url <chr> "https://api.spotify.com/v1/audio-analysi~
## $ time_signature <int> 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 1, 4, 4, 4,~
## $ artists <list> [<data.frame[1 x 6]>], [<data.frame[1 x ~
## $ available_markets <list> <"AD", "AE", "AG", "AL", "AM", "AO", "AR~
## $ disc_number <int> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,~
## $ duration_ms <int> 113217, 167759, 398454, 222670, 211680, 2~
## $ explicit <lgl> TRUE, TRUE, TRUE, TRUE, TRUE, TRUE, TRUE,~
## $ track_href <chr> "https://api.spotify.com/v1/tracks/2EFqMC~
## $ is_local <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE,~
## $ track_name <chr> "Inside Outside", "Here We Go", "Friends ~
## $ track_preview_url <chr> "https://p.scdn.co/mp3-preview/64db1cbde4~
## $ track_number <int> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13~
## $ type <chr> "track", "track", "track", "track", "trac~
## $ track_uri <chr> "spotify:track:2EFqMCOdTTkcFYHoJH21Jr", "~
## $ external_urls.spotify <chr> "https://open.spotify.com/track/2EFqMCOdT~
## $ album_name <chr> "Faces", "Faces", "Faces", "Faces", "Face~
## $ key_name <chr> "C", "F#", "C#", "F", "F#", "A#", "G#", "~
## $ mode_name <chr> "major", "minor", "major", "major", "mino~
## $ key_mode <chr> "C major", "F# minor", "C# major", "F maj~
## $ album_release_date <date> 2014-05-11, 2014-05-11, 2014-05-11, 2014~Wow! That’s a lot of info. To make things a little simpler, we can refer to Spotify’s API documentation to get a better idea of what these variables represent. Some variables are rather self-explanatory, such as artist_name, track_name, album_name, and duration_ms. For this analysis, we want to keep identifying information such as a tracks name, the album it came from, and the release date. The other variables we’ll want to keep are measures about the songs sonic signature. These measures include danceability, energy, loudness, speechiness, acousticness, instrumentalness, liveness, and valence. These are metrics provided by Spotify that give quantitative measures of a track’s audio characteristics. More information on these metrics and how they are derived can be found at Spotify’s API documentation. Other variables we’ll want to keep are key, mode, tempo, time_signature, duration_ms, and explicit. These variables provide more information about the composition of the song.
Now that we’ve identified the variables we want to keep in our working data, we can go ahead and create a filtered version of the full data to move forward with. We can also take this opportunity to reorder our variables into more organized groups.
df <- mm_data %>% select(
# identifying information
track_name,
artist_name,
album_name,
album_release_date,
# spotify provided quantitative measures
acousticness,
danceability,
energy,
instrumentalness,
liveness,
loudness,
speechiness,
valence,
# composition information
duration_ms,
explicit,
key,
mode,
tempo,
time_signature
)# quickly formatting key, mode, and time signature as factors
df <- df %>% mutate(across(c(key, mode, time_signature), as.factor))To keep these posts from getting too long, I’ve decided to break up the project into sections. Let’s recap what we did in this first section: we authenticated our program with the Spotify API, we retrieved the data for the artist we wanted, we corrected some data using an external source, and we formatted the data into a workable form for analysis and clustering. That’s quite a lot in just the first section!
Section two will cover the exploratory data analysis, or EDA, where we’ll take a look at the data through visualizations and draw insights from various plots. However, before we go, let’s remember to save our dataframe that we created so we don’t have to go through these steps again!
write_csv(df, "working-data.csv")Thanks for reading!