🎵 Spotify AI Platform

An End-to-End Machine Learning Platform for Music Mood Analysis, Skip Prediction, and Song Recommendation

---

Live Demo

https://spotify-ai-platform.streamlit.app

📌 Project Overview

Spotify AI Platform is a full-stack Machine Learning application that combines:

• 🎵 Mood-based playlist generation
• ⏭️ Song skip prediction
• ❤️ Music recommendation system
• 📊 Interactive analytics dashboard

The project demonstrates the implementation of:

• Unsupervised Learning
• Supervised Learning
• Recommendation Systems
• Feature Engineering
• Interactive Web Applications

---

🚀 Features

🎵 Mood Playlist Generator

Generate playlists using AI-based mood clustering.

Features:

• Happy
• Relax
• Party
• Workout
• Focus
• Sad

---

⏭️ Skip Prediction

Predict whether users are likely to skip a song.

Algorithm:

• Random Forest Classifier

Input Features:

• Popularity
• Energy
• Danceability
• Tempo
• Acousticness
• Duration
• Mood Cluster

---

❤️ Recommendation Engine

Generate similar song recommendations using:

• Nearest Neighbors
• Cosine Similarity

Output:

• Top 10 similar songs
• Artist information
• Genre information
• Predicted mood

---

📊 Analytics Dashboard

Visualize:

• Genre distributions
• Mood distributions
• Dataset statistics
• Model performance

---

🏗️ Project Architecture

Spotify Dataset
        |
        ▼
Data Cleaning
        |
        ▼
Feature Engineering
        |
        ├────────────► Mood Clustering
        │                  |
        │                  ▼
        │           Playlist Generator
        |
        ├────────────► Skip Prediction
        │                  |
        │                  ▼
        │           Random Forest
        |
        └────────────► Recommendation Engine
                           |
                           ▼
                    Nearest Neighbors
                           |
                           ▼
                      Streamlit App

---

🧠 Machine Learning Pipeline

1. Data Preprocessing

• Missing value handling
• Duplicate removal
• Feature engineering
• Mood generation
• Skip label generation

---

2. Mood Clustering

Algorithm:

K-Means Clustering

Features:

• Danceability
• Energy
• Valence
• Tempo
• Acousticness
• Liveness
• Speechiness
• Instrumentalness

Performance:

Silhouette Score: 0.20
Clusters: 6

---

3. Skip Prediction

Algorithm:

Random Forest Classifier

Performance:

Metric	Score
Accuracy	99%
ROC AUC	0.999
Precision	99%
Recall	98%

---

4. Recommendation System

Algorithm:

Nearest Neighbors

Similarity Metric:

Cosine Similarity

Output:

Top 10 recommendations

---

📈 Feature Importance

Feature	Importance
Duration	0.449
Popularity	0.230
Energy	0.138
Danceability	0.066
Acousticness	0.033

---

🖥️ Application Screenshots

Dashboard

---

Mood Playlist Generator

---

Skip Prediction

---

Recommendation System

---

🛠️ Technologies Used

• Python
• Pandas
• NumPy
• Scikit-learn
• Streamlit
• Plotly
• Matplotlib
• Joblib

---

📂 Project Structure

spotify-ai-platform/

├── datasets/
├── models/
├── notebooks/
├── streamlit_app/
├── assets/
├── reports/
├── requirements.txt
├── README.md
└── LICENSE

---

⚙️ Installation

Clone repository:

git clone https://github.com/Johan621/spotify-ai-platform.git

Install dependencies:

pip install -r requirements.txt

Run application:

cd streamlit_app

streamlit run app.py

---

🎯 Future Improvements

• Spotify API integration
• Deep Learning recommendation system
• User authentication
• Playlist export
• Real-time prediction

---

👨‍💻 Author

Developed an end-to-end Machine Learning project demonstrating:

• Machine Learning
• Recommendation Systems
• Data Science
• Model Explainability
• Interactive AI Applications