Song Classification and Recommendation System

This project focuses on enhancing a raw dataset of songs by extracting detailed musical features, popularity metrics, and genre information using Spotify's URI data. The enriched dataset facilitates deeper analysis and the development of a song recommendation system based on various attributes of the songs.

Table of Contents

• Project Overview
• Setup and Installation
• Data Flow Diagram
• Feature Extraction
• Data Processing
• Recommendation Algorithm
• Usage
• Contributing
• License

Project Overview

The project automates the process of fetching detailed song features using the Spotify API and integrates these features into the original dataset. The key steps include:

• Initial data loading from a CSV file.
• Parallel feature extraction from Spotify URIs.
• Data cleaning and preprocessing.
• Sentiment analysis on song titles.
• Transformation and scaling of features.
• Generation of song recommendations based on user's playlist.

Setup and Installation

# Clone the repository
git clone https://github.com/your-username/song-feature-extraction.git

# Navigate to the project directory
cd song-feature-extraction

# Install dependencies (Assuming Python 3 and pip are installed)
pip install -r requirements.txt

Data Flow Diagram

graph LR
    A[Raw Dataset] -->|Extract URIs| B(URI to Features)
    B --> C{Parallel Processing}
    C -->|Feature Extraction| D[Enriched Dataset]
    D --> E[Data Preprocessing]
    E --> F[Sentiment Analysis]
    F --> G[Feature Transformation & Scaling]
    G --> H[Feature Set Creation]
    H --> I[Generate Recommendations]

Loading

Feature Extraction

• Utilizes uri.py script to extract features from Spotify URIs.
• Features include musical attributes, popularity scores, and genres.
• Implements parallel processing for efficient data handling.

Data Processing

• Cleans and preprocesses the data for analysis.
• Applies sentiment analysis to track names.
• Transforms genre data using TF-IDF.
• Scales numerical features for uniformity.

Recommendation Algorithm

• Summarizes user playlists into a feature vector.
• Calculates cosine similarity between playlist vector and non-playlist songs.
• Recommends songs with the highest similarity scores.

Usage

This project is structured around a Jupyter Notebook, which provides a detailed walkthrough of the data processing, feature extraction, and recommendation generation steps. To use this notebook:

1. Ensure you have Jupyter installed. If not, you can install it using pip:

pip install notebook

2. Start the Jupyter Notebook server from the command line:

jupyter notebook

3. Navigate to the project directory in the Jupyter Notebook web interface.

4. Open the Song_Classification_and_Recommendation_System.ipynb notebook.

5. Run the cells sequentially to perform data loading, feature extraction, data processing, and to generate recommendations.

• Each cell in the notebook is annotated with comments to guide you through the process and explain the purpose of each code block.
• To execute a cell, select it and press Shift + Enter.

Note:

• Make sure all the required dependencies are installed (refer to the Setup and Installation section).
• Ensure you have access to the Spotify API and the necessary API keys if the feature extraction script requires it.

License

This project is licensed under the Apache License - see the LICENSE.md file for details.