Abstract:
Finding individualized and interesting music experiences has grown more difficult in a
time when a wide ocean of musical stuff is easily accessible. This work analyzes the field
of music recommendation using a web-based platform and modern machine learning
methods. Including 100,000 entries and 20 attributes, the dataset combines data from
various sources, such as the 'Spotify Million Song Dataset' and KKBox's Music
Recommendation Challenge datasets. To create a strong recommendation system, the study
uses machine learning algorithms like "CatBoost Classifier," "XGBoost Classifier,"
"LightGBM Classifier," "Random Forest Classifier," and "Extra Trees Classifier."
Handling missing values, feature extraction, and categorization encoding are all part of the
data preprocessing process. The feature engineering process is followed by an exploratory
data analysis (EDA), which offers understanding into the dataset's characteristics. After
extensive development, Extra Trees Classifier is the most successful algorithm,
outperforming the others with an accuracy of 84.63%.The web-based interface, which was
created with Streamlit and Python, easily connects with the Spotify API to offer users a
customized and collaborative music discovery experience. The project follows modern
standards of responsible technology development through placing a high priority on
sustainability, user privacy, and ethical considerations in addition to accuracy
