"Can you write a hit song" Documentation

By Andrew Nickerson, Xavier Evans, and Chris Barber

Note: Since our code and data is rather space-heavy, we have kept it in GitHub for TF grading. If needed, we can resubmit our code in the .zip format.

Live Demo on GitHub Pages Video Demo (Google Drive)

Running Locally

To run the code, create a local server to run index.html after downloading a copy of the entire repository. All JavaScript and other files are included in tags within this HTML file.

File Structure

• index.html contains the main HTML code used to power the web application
• main.js contains the main JavaScript code used
• Other JavaScript files within /js contain separate class code for each visualization, as well as some auxillary script code for other functions (such as searching)
• /audio contains 15 audio files used in the sentiment analysis visualization
• Our data is contained in two directories: /data and /data_scrape. /data_scrape contains data that was scraped using APIs such as the Genius Lyrics API and Spotify API, while /data contains mainly CSV files aggregating the scraped data used in our final website
• CSS is contained in the /css directory
• /song_data contains JSON versions of the data, with some additional fields used in some visualizations
• Other .ipynb notebooks were used for data scraping and cleaning. These files have been included in the repository.

Data

Although there are many variations of our data for organization's sake, the following data fields are held for each song/data point:

• song: the title of the song
• artist: the name(s) of artist(s) in the song
• url: the Genius Lyric URL (used for scraping lyrics)
• lyrics: a dictionary-like structure of lyrics for the song, with keys being the part of the song (such as Verse 1 or Chorus)
• lyrics_text: a raw version of the lyrics (just words)
• rank: a dictionary where keys are dates of hitting the Billboard charts and keys are the rank of the song at that time
• peak_rank: a dictionary where keys are dates of hitting the Billboard charts and keys are peak rank of the song up to that time
• weeks_on_board: a dictionary where keys are dates of hitting the Billboard charts and keys are the number of weeks the song had been on the Billboard up to that time
• last_week: a dictionary where keys are dates of hitting the Billboard charts and keys are where the song was ranked during the previous week
• genre: a list of genres that describe the song/artist, according to Spotify's API
• decade: the decade of release for the song
• release_date: the date of release for the song
• song_features: a dictionary of various song features tracked by Spotify, such as "danceability" and "acousticness"
• duration_ms: the duration of the song in milliseconds
• tempo: the tempo of the song in beats per minute
• time_signature: the (average) number of beats per measure in the song
• positivity: (only tracked for a subset of songs) positivity scores for partitioned segments of the song's lyrics (between $1$ and $-1$)