Sentiment_Analysis_Youtube

Overview

This project is a part of a group effort to analyze the sentiment of YouTube comments comparing the United States and Great Britain. We aim to explore the differences in sentiment between these two regions, and consider the effect of the comments sentiment to the interaction on a video.

Contents

1. Introduction - Background and objectives of the analysis
2. Data Importing & Cleaning - Loading and preparing the YouTube comments dataset
3. Exploratory Data Analysis - Uncovering insights through detailed data exploration
4. Comment Volume Analysis - Comparing number of comments between US and UK
5. Sentiment Polarity Analysis - Using VADER model to score comment sentiment
6. Emoji Analysis - Extracting and analyzing use of emojis
7. Engagement Analysis - Statistical analysis of likes, dislikes, replies
8. Topic Modeling - Using NMF and LDA to extract discussion topics
9. Interactive Visualizations - Plots generated using Plotly Express and Dash
10. Model Development - Training sentiment classification and engagement prediction models

Datasets

1. USvideos.csv and UScomments.csv - Video metadata and comments from United States
2. GBvideos.csv and GBcomments.csv - Video metadata and comments from United Kingdom

These CSV files contain samples of YouTube data scraped and published on Kaggle.

Technologies Used

1. Python
2. Jupyter Notebook
3. Pandas
4. NLTK
5. Matplotlib
6. Seaborn
7. Deepnote Notebook

Results

The sentiment analysis revealed interesting patterns in the comments from both countries. Further details can be found in the Jupyter Notebook provided in this repository.

Future Work

Improve sentiment analysis accuracy by using more advanced natural language processing techniques. Extend the analysis to include comments from other countries for a broader comparison.

How to Run

1. Clone the repository
2. Install dependencies like pandas, nltk, scikit-learn
3. Run jupyter notebook to start Jupyter
4. Open the youtube_comments_analysis.ipynb notebook
5. Run the cells in order to reproduce the analysis
6. Alternatively, view youtube_comments_analysis.html to see a static HTML export of the executed notebook.

Contributors

1. Aryan Jain
2. Devesh Talreja
3. Micah Billington
4. Rupesh Rangwani

Credits

Thanks to DataSnaek for curating and releasing the YouTube comments dataset on Kaggle.

License

This project is licensed under the MIT License.