YouTube Video Analytics Dashboard

Overview

This project dives deep into YouTube trending video analytics through both static visualization and an interactive dashboard. Using raw youtube.csv data, the analysis explores how video performance metrics (views, likes, comments) relate to time of publication, country, and channel behavior.

The objective was to extract insights that guide content strategy, optimize publishing time, and understand audience engagement patterns across different categories and geographies.

Key Features

• Statistical Analysis & Preprocessing:

  • Outlier detection & treatment using IQR and Box-Cox transformation
  • Normality testing using Kolmogorov-Smirnov test
  • Correlation analysis via Pearson heatmaps

• Data Cleaning & Feature Engineering:

  • Cleaned & transformed 14+ raw features (e.g. tags, publish time, part of day)
  • Generated new attributes like part_of_day, weekday, publish_country, etc.

• Static Visualizations (Matplotlib & Seaborn):

  • Histograms, Boxplots, Heatmaps, Line Charts, Pie Charts, Contour Plots

• Interactive Dashboard (Dash & Plotly):

  • Channel-level metrics: Views, Likes, Comments, Dislikes
  • Time of day vs day-of-week publishing impact
  • Country-wise distribution and trends
  • Filters for year, country, and custom checklists

• Tech Stack:
  Python, Pandas, NumPy, Seaborn, Matplotlib, Plotly, Dash, Scikit-learn, PrettyTable

Skills Demonstrated

• Data wrangling & EDA
• Dimensionality Reduction (PCA)
• Dashboard development
• Data storytelling with plots
• Insights generation for strategic planning

Repository Structure

YouTube-Analytics-Dashboard
├── FTP(G40559527).py                     # Full static analysis + plots
├── Youtube Dashboard.py                  # Dash application for interactivity
├── youtube.csv                           # Dataset
├── Youtube Video Analysis(REPORT).pdf    # Summary report of project
├── README.md                             # Project overview (this file)

Insights Gleaned

• Afternoon videos gain higher engagement across most categories.
• Channels with consistent posting on weekends outperform others in views.
• Certain countries show skewed like/comment ratios indicating regional behavior.

How to Run Locally

# 1. Clone the repository
git clone https://github.com/your-username/youtube-analytics-dashboard.git
cd youtube-analytics-dashboard

# 2. Install requirements
pip install -r requirements.txt

# 3. Run the Dash App
python 'Youtube Dashboard.py'

Let's Connect

Author: Surya Vamsi Patiballa
MS in Data Science, George Washington University

• Email :- svamsi2002@gmail.com
• Resume :- https://drive.google.com/file/d/178IYcArC6YYVdJiIwRmJYodzKZ-JXe-D/view?usp=sharing
• LinkedIn :- https://www.linkedin.com/in/surya-patiballa-b724851aa/