YouTube Trending Video Analysis & Prediction (India 2017–2018)

This project focuses on analyzing and predicting trends in YouTube videos using Data Science and Machine Learning techniques. Using real-world data from India's trending videos in 2017 and 2018, the aim is to understand what makes a video go viral and whether we can accurately predict its performance.

Project Overview

In this project, I performed: Exploratory Data Analysis (EDA) to uncover trends and patterns in the dataset. Feature Engineering to extract meaningful insights from video metrics. Predictive Modeling using: Linear Regression to predict the view count. XGBoost Classifier to determine if a video will trend or not.

Dataset

Source: Kaggle – YouTube Trending Video Dataset Focus: Indian region (INvideos.csv) Features include: title, channel_title, category_id, views, likes, dislikes, comment_count, publish_time, etc.

Technologies Used

• Python
• Pandas, NumPy
• Matplotlib, Seaborn (Visualization)
• Scikit-learn (ML models & preprocessing)
• XGBoost (Boosting algorithm for classification)

Key Goals

• Understand engagement metrics and their influence on video popularity.
• Predict future view counts with Linear Regression.
• Classify whether a video will trend using XGBoost.
• Gain deeper insights into YouTube's trending dynamics.

Future Enhancements

• Build a web application to allow users to input video details or links and get predictions instantly.
• Use YouTube Data API to auto-fetch video stats via URL.
• Improve visualizations and interactivity with dashboards.