Tweet Prediction Analyzer

A Python-based Streamlit web app to extract, analyze, and visualize predictions from tweets using LangExtract.

---

Features

• Extract predictions from tweet datasets (CSV files) using LangExtract.
• View, filter, and validate predictions interactively.
• Visualize prediction statistics and trends.
• LLM verification integration for predictions.
• Export validated predictions.

---

Prerequisites

• Python 3.9+
• Streamlit
• LangExtract Python SDK
• pandas, plotly

---

Installation

1. Clone this repository:

git clone https://github.com/yourusername/tweet-prediction-analyzer.git
cd tweet-prediction-analyzer

2. Install dependencies:

pip install -r requirements.txt

3. Create a .env file in the project root with the following variables:

INPUT_FOLDER=path/to/csv/folder
LANGEXTRACT_API_KEY=your_langextract_api_key

• INPUT_FOLDER: Path to the folder containing your tweet CSV files.
• LANGEXTRACT_API_KEY: API key for LangExtract.

---

CSV File Structure

Your tweet CSV should have the following columns:

id,tweetText,tweetURL,type,tweetAuthor,handle,replyCount,quoteCount,retweetCount,likeCount,views,bookmarkCount,createdAt,allMediaURL,videoURL

• id: Unique tweet identifier
• tweetText: Content of the tweet
• tweetURL: URL of the tweet
• type: Type of tweet (optional)
• tweetAuthor: Name of the author
• handle: Twitter handle
• replyCount, quoteCount, retweetCount, likeCount, views, bookmarkCount: Tweet engagement metrics
• createdAt: Timestamp of the tweet (format YYYY-MM-DD HH:MM:SS)
• allMediaURL, videoURL: Media links

---

Running the App Locally

streamlit run app.py

• Open your browser at http://localhost:8501.
• Use the sidebar to manage data, run extraction, and view analytics.

---

Project Structure

tweet-prediction-analyzer/
│
├── app.py                  # Main Streamlit app
├── main.py                 # TweetPredictionAnalyzer class and extraction logic
├── requirements.txt        # Python dependencies
├── .env                    # Environment variables
├── README.md               # Documentation
├── outputs/                # Folder for predictions and JSON files
└── data/                   # CSV files (tweets)

---

License

MIT License