jhunhwang / Music_Genre_Classification

Music Genre Classification

Music_Genre_Classification (K-POP VS Western)

A complete machine learning pipeline from scraping music data to modelling a music genre classifier and deployting it.

The deployed web app is live at https://jihun-kpop-western-classifier.herokuapp.com/ .

Project Scope

• Data Collection (Spotify API for Music Features and Genius API for Song Lyrics)

• Data Cleaning (Treating Null Values Inside the Dataset, etc.)

• Exploratory Data Analysis

• Machine Learning Modelling and Testing (Classification Model)

• Model Deployment

• Technologies Used

• Data Collection & Cleaning

  • Song Features
    • Spotipy API (For Collecting Music Features)
    • Pandas (Inserting the data into a dataframe, Dropping Duplicate Rows and Exporting the data as CSV)
  • Song Lyrics
    • Requests (Send request to Genius API for song info)
    • Googlesearch (If requests is unable to get the song info, a google search will be performed to find the link of the music in Genius.)
    • BeautifulSoup (With the URLs returned from either Requests or Googlesearch, BS will scrape the lyrics of the song.)
    • Pandas (Used to insert the lyrics into a dataframe)

• Exploratory Data Analysis

  • Pandas (Analyzing, Filtering, Summary)
  • Matplotlib, Seaborn (Visualization)
  • Statsmodels (ANOVA)
  • Scipy (Mannwhitneyu)
  • Itertools (For different Group Combinations)

• Machine Learning Modelling

  • Sci-Kit Learn
  • XGBoost
  • Light GBM

• Model Deployment

  • Streamlit