Skip to content

Latest commit

 

History

History
110 lines (81 loc) · 3.03 KB

README.md

File metadata and controls

110 lines (81 loc) · 3.03 KB
Raw

Financial News Sentiment Analyzer

Overview

This Python script analyzes the sentiment of financial news articles using the Google News API and OpenAI's GPT-3.5 model. It fetches articles based on a specified topic, analyzes their sentiment, and provides a summary of the overall sentiment trends.

Features

  • Fetch news articles from Google News based on a specified topic
  • Analyze sentiment of articles using OpenAI's GPT-3.5 model
  • Cache sentiment analysis results to avoid redundant API calls
  • Provide summary analysis of sentiment trends
  • Output results to CSV file (optional)
  • Command-line interface for easy usage and customization

Requirements

  • Python 3.6+
  • OpenAI API key (set as environment variable OPENAI_API_KEY)
  • Required Python packages (install via pip install -r requirements.txt):
    • GoogleNews
    • newspaper3k
    • tqdm
    • openai

Installation

  1. Clone the repository:

    git clone https://github.com/jmcdice/newsfeel.git
cd newsfeel

  2. Install the required packages:

    python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

  3. Set up your OpenAI API key as an environment variable:

    export OPENAI_API_KEY='your-api-key-here'

Usage

Run the script using the following command:

python newsfeel.py [options]

Options:

  • -n, --num_articles: Number of articles to process (default: 5)
  • --print_cache: Print everything in the cache
  • --analyze_cache: Analyze cache sentiments and exit
  • --analyze_summaries: Analyze summaries of cached articles and exit
  • -t, --topic: Topic to fetch news for (default: "Financial News")
  • --loglevel: Logging level (DEBUG, INFO, WARNING, ERROR, CRITICAL)
  • -o, --output_file: File to write output results to (CSV format)

Examples:

  1. Analyze 10 articles about "Bitcoin":

    python newsfeel.py -n 10 -t "Bitcoin"

  2. Analyze cache sentiments for previously fetched "Stock Market" articles:

    python newsfeel.py --analyze_cache -t "Stock Market"

  3. Generate a summary analysis of cached "Cryptocurrency" articles:

    python newsfeel.py --analyze_summaries -t "Cryptocurrency"

  4. Output results to a CSV file:

    python newsfeel.py -n 20 -t "Gold Price" -o results.csv

Output

The script provides the following output:

  1. Sentiment analysis for each processed article
  2. Overall sentiment analysis summary, including:
    • General sentiment (Very Bearish, Bearish, Neutral, Bullish, Very Bullish)
    • Total number of articles analyzed
    • Average confidence score
    • Sentiment distribution
    • Weighted sentiment score

Caching

The script uses a caching mechanism to store sentiment analysis results. This helps to:

  • Reduce API calls to OpenAI
  • Speed up subsequent runs on the same topic
  • Allow for offline analysis of previously fetched articles

Cache files are stored in the cache directory with filenames based on the analyzed topic.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.