NFL Twitter Sentiment Analysis

NFL Team Sentiment Analysis Dataset

Overview

This dataset contains sentiment-labeled social media posts about NFL teams, specifically focused on tracking fan reactions and public sentiment. Each entry has been processed through a fine-tuned RoBERTa model with custom sports context modifiers to ensure accurate sentiment classification in the context of NFL-specific language and events.

Data was scraped via selenium web scraper
📝 View Code on GitHub

Dataset Description

Features

• username: Author of the post
• timestamp: Time of post creation (format: YYYY-MM-DDThh:mm:ss.000Z)
• text: Content of the post
• team: NFL team referenced in the post
• sentiment: Classified sentiment (positive, negative, or neutral)
• confidence: Model's confidence score for the sentiment classification (0-1)
• roberta_raw_outputs: Raw probability distributions from the model [negative, neutral, positive]

Key Statistics

• Total Records: 3701 labeled posts
• Time Period: November 2024
• Teams Covered: Denver Broncos, Indianapolis Colts, Cleveland Browns, Jacksonville Jaguars, Kansas City Chiefs, Detroit Lions

Collection Methodology

• Data sourced from public social media posts on twitter
• Processed through a RoBERTa model fine-tuned for NFL context
• Custom sports-specific context modifiers applied for improved accuracy
• Automated cleaning to remove sales-related posts, media-only content, and irrelevant material

Potential Applications

• Fan sentiment analysis during NFL season
• Team performance impact on public opinion
• Temporal analysis of fan reactions
• Sports marketing research
• Social media engagement studies

Limitations

Does not currently have all of the NFL teams, but the dataset will be updated over time and I will soon remove this.

License

MIT

Note: This dataset contains publicly available social media posts. Users should comply with platform terms of service and privacy considerations.