🎬 Netflix Content Engagement Prediction - End-to-End ML Pipeline

Author: Utkarsh Panchal
Platform: Google Colab + AWS SageMaker
Repository Type: Full-stack ML Pipeline (data → model → insights)
Tech Stack: Python · pandas · NumPy · scikit-learn · XGBoost · AWS SageMaker · boto3 · Matplotlib · Seaborn

---

🧠 Overview

This project builds an end-to-end machine-learning pipeline that predicts how users engage with Netflix titles.
The model analyzes content metadata (genres, ratings, duration, etc.) to forecast:

• ⭐ User Rating (1–5) — Regression
• ❤️ User Liked / Not Liked (0 / 1) — Classification

It demonstrates how streaming platforms can leverage ML to improve recommendations, catalog decisions, and user retention.

---

🎯 Business Motivation

Streaming companies like Netflix depend on predicting viewer engagement before a title is released.
Accurate engagement prediction enables teams to:

• Prioritize which content to promote or license
• Personalize recommendations for users
• Allocate production budgets strategically
• Reduce churn through smarter suggestions

This notebook turns raw catalog data → actionable engagement insights.

---

⚙️ Technical Architecture

Phase	Goal	Tools
1️⃣ Data Preparation	Cleaning, feature engineering, encoding	pandas · NumPy
2️⃣ Model Training	Regression + Classification (XGBoost)	scikit-learn · XGBoost
3️⃣ Evaluation	Compute RMSE, R², Accuracy, Precision, Recall, F1	sklearn.metrics
4️⃣ Visualization	Genre + prediction analysis	Matplotlib · Seaborn
5️⃣ Deployment (Optional)	Realtime endpoint hosting	AWS SageMaker
6️⃣ Monitoring	Latency + drift tracking	AWS CloudWatch

---

🧩 Pipeline Flow

Raw Netflix Data
     ↓
Data Cleaning & Feature Engineering
     ↓
Train/Test Split (80/20)
     ↓
Model Training (XGBoost Regression + Classification)
     ↓
Model Evaluation & Visualization
     ↓
[Optional] AWS SageMaker Deployment