📍 Overview

This project analyzes a dataset of Spotify tracks to identify distinct clusters of songs based on their audio characteristics (e.g., danceability, energy, acousticness).

By applying Principal Component Analysis (PCA) for dimensionality reduction and K-Means Clustering, we segment songs into groups. These clusters are then used to power a Content-Based Recommendation System that suggests songs mathematically similar to a user's input.

Key Features

🧠 Methodology

The project follows a standard Data Science pipeline:

1. Data Preprocessing: * Removal of duplicates and missing values.
   • Dropping non-numeric identifiers (track_id, track_uri, etc.).
2. Feature Standardization: * Uses StandardScaler to normalize features (Mean=0, Std=1) so that high-magnitude features (like Loudness) don't dominate the distance calculations.
3. Dimensionality Reduction:
   • PCA (Principal Component Analysis) is applied to compress the 9 audio features into 2 primary components for visualization purposes.
4. Clustering:
   • K-Means Algorithm groups the data into 5 clusters (k=5).
   • $$\text{Minimize } \sum_{i=1}^{k} \sum_{x \in S_i} || x - \mu_i ||^2$$

📐 System Architecture

flowchart LR
    subgraph Input
    A[Raw Dataset] --> B(Data Cleaning)
    end

    subgraph Feature Engineering
    B --> C[Feature Selection]
    C -->|Danceability, Energy, etc.| D[StandardScaler]
    end

    subgraph Modeling
    D --> E[PCA Reduction]
    D --> F[K-Means Clustering]
    E --> G[2D Visualization]
    F --> H[Cluster Labels]
    end

    subgraph Output
    H --> I{Recommendation Engine}
    I -->|Input Song| J[Output Top 5 Similar Songs]
    end

Project Structure Plaintext

├── spotify_analysis.ipynb # Main Jupyter Notebook (or .py script) containing all logic

├── spotify dataset.csv # Source data (ensure this is in root)

├── requirements.txt # Python dependencies

└── README.md # Project documentation

Installation & Usage

1. Clone the Repository Bash git clone https://github.com/your-username/spotify-genre-segmentation.git cd spotify-genre-segmentation
2. Install Dependencies Bash pip install -r requirements.txt
3. Run the Analysis If using the Python script:

Bash python spotify_analysis.py 4. Get Recommendations To use the recommendation function inside the script/notebook:

Python

Example Usage

rec = recommend_songs('Shape of You', df_final) print(rec) 📊 Visualizations The script generates several key insights:

Feature Distributions: Histograms showing the spread of Audio Features.

Correlation Matrix: A heatmap displaying how features like Energy and Loudness are positively correlated, while Acousticness and Energy are negatively correlated.

PCA Cluster Plot: A scatter plot visualizing how the K-Means algorithm has grouped the songs in 2D space.