DEV Community

Cover image for End-to-End YouTube Channel Analytics Pipeline
Lagat Josiah
Lagat Josiah

Posted on • Edited on

End-to-End YouTube Channel Analytics Pipeline

#docker #kafka #pyspark #grafana

End-to-End YouTube Channel Analytics Pipeline

Complete Implementation Guide

Student Name: Lagat Josiah
Date: October 10, 2025

Channel Analyzed: Channel Name

Table of Contents

  1. Executive Summary
  2. Project Architecture
  3. Prerequisites & Environment Setup
  4. Phase 1: Data Ingestion
  5. Phase 2: Data Processing with PySpark
  6. Phase 3: Workflow Orchestration with Airflow
  7. Phase 4: Visualization with Grafana
  8. Phase 5: Containerization with Docker
  9. Results & Insights
  10. Challenges & Solutions
  11. Conclusion & Future Work

Executive Summary

This project demonstrates the implementation of a complete data engineering pipeline for YouTube channel analytics. The solution ingests data from YouTube Data API v3, processes it using Apache Spark, orchestrates workflows with Apache Airflow, visualizes insights through Grafana dashboards, and packages everything using Docker containers.

Key Technologies:

  • Data Ingestion: YouTube Data API v3, Python
  • Data Processing: Apache Spark (PySpark)
  • Orchestration: Apache Airflow 3.0.0
  • Visualization: Grafana
  • Storage: Aiven PostgreSQL (Cloud)
  • Containerization: Docker & Docker Compose

Project Architecture 

┌─────────────────────────────────────────────────────────────┐
│                    YouTube Data API v3                      │
└────────────────────┬────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│              Data Ingestion Layer (Python)                   │
│  • Fetch channel statistics                                  │
│  • Fetch video metadata & metrics                            │
│  • Store raw data as JSON                                    │
└────────────────────┬────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│           Processing Layer (Apache Spark)                    │
│  • Load JSON data into DataFrames                            │
│  • Transform timestamps                                      │
│  • Calculate engagement metrics                              │
│  • Feature engineering                                       │
└────────────────────┬────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│              Storage Layer (Aiven PostgreSQL)                │
│  • channel_stats table                                       │
│  • video_stats table                                         │
└────────────────────┬────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│          Orchestration Layer (Apache Airflow)                │
│  DAG: Extract → Transform → Load → Validate                  │
│  Schedule: Daily execution                                   │
└─────────────────────────────────────────────────────────────┘
                     │
                     ▼
┌─────────────────────────────────────────────────────────────┐
│           Visualization Layer (Grafana)                      │
│  • Top videos dashboard                                      │
│  • Engagement metrics                                        │
│  • Publishing patterns analysis                              │
└─────────────────────────────────────────────────────────────┘
Enter fullscreen mode Exit fullscreen mode

Prerequisites & Environment Setup

System Requirements

  • Operating System: WSL2 (Ubuntu 20.04+) on Windows
  • RAM: Minimum 4GB (8GB recommended)
  • Storage: 10GB free space
  • Software:
    • Docker Engine 20.10+
    • Docker Compose v2+
    • Python 3.11+
    • Java JDK 17 (for PySpark)

Initial Setup Commands 

# Update system
sudo apt update && sudo apt upgrade -y

# Install Docker
sudo apt install docker.io -y
sudo usermod -aG docker $USER
newgrp docker

# Verify installations
docker --version
python3 --version
Enter fullscreen mode Exit fullscreen mode

Project Structure 

youtube-analytics-pipeline/
├── dags/
│   └── youtube_analytics_dag.py
├── scripts/
│   ├── youtube_ingestion.py
│   └── spark_processing.py
├── data/
│   ├── raw/
│   └── processed/
├── config/
├── logs/
├── plugins/
├── docker-compose.yaml
├── Dockerfile
├── requirements.txt
└── .env
Enter fullscreen mode Exit fullscreen mode

Phase 1: Data Ingestion

1.1 YouTube API Setup

Steps to obtain API credentials:

  1. Navigate to Google Cloud Console
  2. Create a new project: "YouTube Analytics Pipeline"
  3. Enable "YouTube Data API v3"
  4. Create API credentials (API Key)
  5. Note the API key securely

1.2 Channel Selection

Selected Channel Details:

  • Channel ID: UC_x5XG1OV2P6uZZ5FSM9Ttw
  • Rationale: [Explain why you chose this channel]

1.3 Data Ingestion Implementation

File: scripts/youtube_ingestion.py

Key Features:

  • Fetches channel-level statistics (subscribers, total views, video count)
  • Retrieves video-level data (title, publish date, views, likes, comments)
  • Implements pagination for large datasets
  • Saves raw data as JSON with timestamps

Code Structure:

def fetch_channel_data(api_key, channel_id):
    """Fetch channel statistics from YouTube API"""
    # Implementation details in source file

def fetch_videos_data(api_key, channel_id, max_results=50):
    """Fetch video data with pagination"""
    # Implementation details in source file

def save_data(data, filename):
    """Save data to JSON file with timestamp"""
    # Implementation details in source file
Enter fullscreen mode Exit fullscreen mode

Sample Output:

data/raw/
├── channel_data_20251010_120000.json
└── videos_data_20251010_120000.json
Enter fullscreen mode Exit fullscreen mode

Phase 2: Data Processing with PySpark

2.1 Spark Session Configuration

File: scripts/spark_processing.py

Configuration:

  • Application Name: "YouTubeAnalytics"
  • JDBC Driver: PostgreSQL JDBC 42.7.1
  • SSL enabled for secure connections

2.2 Data Transformations

Channel Data Processing:

  • Extract channel metadata
  • Parse statistics (subscribers, views, video count)
  • Add processing timestamp

Video Data Processing:

  1. Timestamp Conversion:

    • Convert ISO 8601 timestamps to datetime objects
    • Extract hour and day of week for analysis

  2. Feature Engineering:

    • Calculate engagement rate: (likes + comments) / views * 100
    • Handle missing values (default to 0)
    • Parse video duration

  3. Data Quality:

    • Remove null values
    • Validate numeric fields
    • Ensure data consistency

2.3 Data Schema

channel_stats table:

├── channel_id (string)
├── channel_name (string)
├── subscribers (integer)
├── total_views (integer)
├── video_count (integer)
└── fetch_date (timestamp)
Enter fullscreen mode Exit fullscreen mode

video_stats table:

├── video_id (string)
├── title (string)
├── published_at (string)
├── published_datetime (timestamp)
├── publish_hour (integer)
├── publish_day (integer)
├── views (integer)
├── likes (integer)
├── comments (integer)
├── duration (string)
└── engagement_rate (double)
Enter fullscreen mode Exit fullscreen mode

2.4 Storage Strategy

Primary Storage: Aiven PostgreSQL (Cloud)

  • Host: pg-c63647-lagatkjosiah-692c.c.aivencloud.com
  • Port: 24862
  • Database: defaultdb
  • SSL Mode: Required

Backup Storage: Parquet files

  • Location: data/processed/
  • Format: Apache Parquet (columnar storage)

Phase 3: Workflow Orchestration with Airflow

3.1 Airflow Architecture

Executor: LocalExecutor (simplified for development)

Components:

  • API Server (port 8080)
  • Scheduler
  • DAG Processor
  • Triggerer

3.2 DAG Implementation

File: dags/youtube_analytics_dag.py

DAG Configuration:

default_args = {
    'owner': 'airflow',
    'depends_on_past': False,
    'start_date': datetime(2025, 1, 1),
    'retries': 1,
    'retry_delay': timedelta(minutes=5),
}
Enter fullscreen mode Exit fullscreen mode

Schedule: Daily execution (@daily)

3.3 Task Pipeline 

┌─────────────────┐
│  Extract Task   │  ← Fetch data from YouTube API
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ Transform Task  │  ← Process data with PySpark
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ Validate Task   │  ← Verify data integrity
└─────────────────┘
Enter fullscreen mode Exit fullscreen mode

Task Definitions:

  1. extract_youtube_data:

    • Calls YouTube API
    • Saves raw JSON
    • Returns timestamp

  2. transform_with_pyspark:

    • Loads raw data
    • Applies transformations
    • Saves to PostgreSQL

  3. validate_data:

    • Confirms successful execution
    • Logs completion status

3.4 Monitoring & Logging

Access: http://localhost:8080

  • Username: airflow
  • Password: airflow

Features:

  • Visual DAG graph
  • Task execution logs
  • Retry mechanisms
  • Execution history

Phase 4: Visualization with Grafana

4.1 Data Source Configuration

Connection Details:

Name: Aiven YouTube Analytics
Type: PostgreSQL
Host: pg-c63647-lagatkjosiah-692c.c.aivencloud.com:24862
Database: defaultdb
User: avnadmin
SSL Mode: require
Enter fullscreen mode Exit fullscreen mode

4.2 Dashboard 1: Top Performing Videos

Query:

SELECT 
  title,
  views,
  likes,
  comments,
  engagement_rate
FROM video_stats
ORDER BY views DESC
LIMIT 10
Enter fullscreen mode Exit fullscreen mode

Visualization: Bar Chart

  • X-axis: Video Title
  • Y-axis: View Count
  • Color: Engagement Rate (gradient)

Insights:

  • Identifies highest performing content
  • Correlates views with engagement
  • Helps understand audience preferences

4.3 Dashboard 2: Engagement Rate Trends

Query:

SELECT 
  DATE(published_datetime) as date,
  AVG(engagement_rate) as avg_engagement,
  SUM(views) as total_views,
  COUNT(*) as video_count
FROM video_stats
GROUP BY DATE(published_datetime)
ORDER BY date DESC
LIMIT 30
Enter fullscreen mode Exit fullscreen mode

Visualization: Time Series Graph

  • X-axis: Date
  • Y-axis: Average Engagement Rate
  • Secondary Y-axis: Total Views

Insights:

  • Tracks engagement over time
  • Identifies trends and patterns
  • Reveals optimal posting times

4.4 Dashboard 3: Publishing Time Analysis

Query:

SELECT 
  publish_day,
  publish_hour,
  COUNT(*) as video_count,
  AVG(views) as avg_views,
  AVG(engagement_rate) as avg_engagement
FROM video_stats
GROUP BY publish_day, publish_hour
ORDER BY publish_day, publish_hour
Enter fullscreen mode Exit fullscreen mode

Visualization: Heatmap

  • X-axis: Hour of Day (0-23)
  • Y-axis: Day of Week (1-7)
  • Color Intensity: Average Views or Engagement

Insights:

  • Optimal publishing schedule
  • Audience activity patterns
  • Best times for maximum reach

4.5 Dashboard 4: Channel Growth Metrics

Query:

SELECT 
  fetch_date,
  subscribers,
  total_views,
  video_count
FROM channel_stats
ORDER BY fetch_date DESC
Enter fullscreen mode Exit fullscreen mode

Visualization: Multi-line Graph

Insights:

  • Subscriber growth trajectory
  • View accumulation rate
  • Content production velocity

Phase 5: Containerization with Docker

5.1 Dockerfile

Base Image: apache/airflow:3.0.0

Key Components:

# Install Java 17 for PySpark
RUN apt-get install -y openjdk-17-jdk-headless

# Install Python dependencies
RUN pip install --no-cache-dir -r /requirements.txt

# Download PostgreSQL JDBC driver
RUN curl -L https://jdbc.postgresql.org/download/postgresql-42.7.1.jar
Enter fullscreen mode Exit fullscreen mode

5.2 Docker Compose Services

Services Architecture:

  1. postgres - Airflow metadata database
  2. redis - Message broker for Celery
  3. airflow-apiserver - Web UI (port 8080)
  4. airflow-scheduler - Task scheduler
  5. airflow-dag-processor - DAG parser
  6. airflow-triggerer - Event-driven tasks
  7. grafana - Visualization (port 3000)

5.3 Volume Mounts 

volumes:
  - ./dags:/opt/airflow/dags
  - ./logs:/opt/airflow/logs
  - ./scripts:/opt/airflow/scripts
  - ./data:/opt/airflow/data
  - ./config:/opt/airflow/config
  - ./plugins:/opt/airflow/plugins
Enter fullscreen mode Exit fullscreen mode

5.4 Environment Variables

Configuration via .env file:

YOUTUBE_API_KEY=<your_api_key>
YOUTUBE_CHANNEL_ID=UC_x5XG1OV2P6uZZ5FSM9Ttw
POSTGRES_USER=avnadmin
POSTGRES_PASSWORD=<password>
POSTGRES_HOST=pg-c63647-lagatkjosiah-692c.c.aivencloud.com
POSTGRES_PORT=24862
POSTGRES_DB=defaultdb
POSTGRES_SSL_MODE=require
Enter fullscreen mode Exit fullscreen mode

5.5 Deployment Commands 

# Build custom image
docker compose build

# Start all services
docker compose up -d

# Check service health
docker compose ps

# View logs
docker compose logs -f

# Stop services
docker compose down
Enter fullscreen mode Exit fullscreen mode

Results & Insights

Key Findings

1. Content Performance Analysis

Top 3 Videos by Views:

  1. [Video Title] - [Views] views - [Engagement Rate]%
  2. [Video Title] - [Views] views - [Engagement Rate]%
  3. [Video Title] - [Views] views - [Engagement Rate]%

Observations:

  • [Describe patterns in top-performing content]
  • [Common characteristics of successful videos]

2. Engagement Metrics

Average Engagement Rate: [X]%

  • Likes per 1000 views: [X]
  • Comments per 1000 views: [X]
  • Engagement trend: [Increasing/Decreasing/Stable]

3. Publishing Strategy Insights

Optimal Publishing Times:

  • Best Day: [Day of week]
  • Best Hour: [Hour range]
  • Peak Engagement Window: [Time range]

Content Frequency:

  • Average videos per week: [X]
  • Most productive month: [Month]

4. Channel Growth Trajectory

Growth Metrics (Last 30 days):

  • Subscriber growth: [+X%]
  • View growth: [+X%]
  • Content output: [X videos]

Challenges & Solutions

Challenge 1: Docker Build Segmentation Fault

Problem: Docker Compose build command failed with segmentation fault

Root Cause: Conflict between BuildKit and legacy builder

Solution:

DOCKER_BUILDKIT=0 docker compose build
Enter fullscreen mode Exit fullscreen mode

Learning: WSL2 environment requires specific Docker configurations

Challenge 2: Celery Worker Not Starting

Problem: Tasks stuck in "queued" state, worker not processing

Root Cause: CeleryExecutor complexity in development environment

Solution: Switched to LocalExecutor

AIRFLOW__CORE__EXECUTOR: LocalExecutor
Enter fullscreen mode Exit fullscreen mode

Learning: Choose appropriate executor based on deployment scale

Challenge 3: Environment Variables Not Loading

Problem: YouTube API credentials not accessible in containers

Root Cause: Variables not passed through docker-compose.yaml

Solution: Added explicit environment mapping

environment:
  YOUTUBE_API_KEY: ${YOUTUBE_API_KEY}
  YOUTUBE_CHANNEL_ID: ${YOUTUBE_CHANNEL_ID}
Enter fullscreen mode Exit fullscreen mode

Learning: Container isolation requires explicit configuration

Challenge 4: PySpark JDBC Driver Missing

Problem: Spark couldn't connect to PostgreSQL

Root Cause: PostgreSQL JDBC driver not included in base image

Solution: Added driver download in Dockerfile

RUN curl -L https://jdbc.postgresql.org/download/postgresql-42.7.1.jar
Enter fullscreen mode Exit fullscreen mode

Learning: Custom images must include all dependencies

Technical Specifications

Performance Metrics

  • Data Ingestion Speed: ~100 videos/minute
  • Spark Processing Time: 15-30 seconds (100 videos)
  • DAG Execution Time: 2-3 minutes (full pipeline)
  • Storage Efficiency: Parquet reduces size by ~70%

Scalability Considerations

  • Current Capacity: 1000+ videos per run
  • API Quota: 10,000 units/day (YouTube API)
  • Database Scaling: Horizontal with Aiven PostgreSQL
  • Container Scaling: Docker Swarm or Kubernetes ready

Security Measures

  • SSL/TLS encryption for all database connections
  • API keys stored in environment variables (not in code)
  • Aiven managed PostgreSQL with automatic backups
  • Docker network isolation between services

Conclusion & Future Work

Project Achievements

✅ Successfully implemented end-to-end data pipeline

✅ Automated daily data collection and processing

✅ Created real-time analytics dashboards

✅ Containerized entire solution for portability

✅ Demonstrated proficiency in modern data engineering tools

Learning Outcomes

  1. Data Engineering: Designed and implemented ETL pipeline
  2. Cloud Integration: Utilized managed database services (Aiven)
  3. Orchestration: Mastered Apache Airflow workflow management
  4. Containerization: Built production-ready Docker setup
  5. Visualization: Created actionable business intelligence dashboards

Future Enhancements

Short-term Improvements

  1. Sentiment Analysis: Add NLP processing for video comments
  2. Competitor Analysis: Compare multiple channels
  3. Alert System: Email notifications for anomalies
  4. API Rate Limiting: Intelligent quota management

Long-term Roadmap

  1. Machine Learning: Predictive models for view forecasting
  2. Real-time Processing: Apache Kafka for streaming data
  3. Advanced Analytics: Audience demographic insights
  4. Mobile Dashboard: Grafana mobile app integration
  5. Multi-cloud Deployment: AWS/Azure/GCP compatibility

Business Value

This pipeline provides:

  • Data-driven decision making for content strategy
  • Automated reporting saving hours of manual work
  • Trend identification for competitive advantage
  • Scalable infrastructure supporting business growth

References & Resources

Documentation

Tools & Technologies

  • Python 3.11
  • Apache Airflow 3.0.0
  • Apache Spark 3.5.0
  • PostgreSQL 13
  • Docker Engine 20.10+
  • Grafana 10.x

Source Code Repository

[Link to GitHub/GitLab repository with complete code]

Appendix

A. Complete File Listings

requirements.txt:

google-api-python-client==2.108.0
pyspark==3.5.0
psycopg2-binary==2.9.9
py4j==0.10.9.7
Enter fullscreen mode Exit fullscreen mode

B. Useful Commands Reference 

# Start pipeline
docker compose up -d

# Stop pipeline
docker compose down

# View logs
docker compose logs -f [service_name]

# Rebuild after changes
docker compose build --no-cache

# Access Airflow CLI
docker compose exec airflow-scheduler airflow dags list

# Trigger DAG manually
docker compose exec airflow-scheduler airflow dags trigger youtube_analytics_pipeline
Enter fullscreen mode Exit fullscreen mode

C. Troubleshooting Guide

Issue Solution
DAG not appearing Check dags/ folder permissions
Task fails Review logs in Airflow UI
Connection timeout Verify network connectivity
Out of memory Increase Docker memory allocation

Presentation Date: 10/10/2025

Duration: 15 minutes

Q&A: 5 minutes

This project demonstrates comprehensive data engineering skills including data ingestion, processing, orchestration, visualization, and containerization using industry-standard tools and best practices.

Top comments (0)