Welcome to the Fake News Detection App! This application leverages machine learning models to help you analyze news articles and classify them as True or Fake. With the power of Natural Language Processing (NLP) and ML Algorithms, this app can assist in combating misinformation and promoting trustworthy news.

• Multi-Model Analysis: Use multiple machine learning models for news classification:

  • Logistic Regression 🧠
  • Decision Tree 🌳
  • Random Forest 🌲
  • Gradient Boosting ⚡

• Real-Time News Classification: Enter a news article, and the app will analyze it in real-time to classify it as True ✅ or Fake ❌.

• Evaluation Metrics: View detailed performance metrics for each model, including accuracy, precision, recall, and F1 score.

• Easy-to-Use Interface: Simply input the news article and click the Analyze button to see results.

• Streamlit: For building the interactive UI.
• scikit-learn: For training and evaluating the machine learning models.
• Pandas: For data manipulation.
• TfidfVectorizer: For text vectorization (converting text into numerical form).
• Joblib: For saving and loading trained models.
• Regular Expressions (Regex): For preprocessing and cleaning the news data.

git clone https://github.com/your-username/fake-news-detection.git
cd fake-news-detection

Ensure you have Python 3.x installed, then create a virtual environment and install the necessary packages:

# Create a virtual environment
python -m venv venv
# Activate the virtual environment
source venv/bin/activate  # For Linux/Mac
venv\Scripts\activate     # For Windows

# Install required packages
pip install -r requirements.txt

streamlit run app.py

Your Fake News Detection app will now be running locally at http://localhost:8501.

• We load two datasets, one containing True news and the other Fake news.
• The text is cleaned using various techniques like:
  • Lowercasing the text.
  • Removing URLs and HTML tags.
  • Removing punctuation and numbers.
  • Stripping newlines and extra spaces.

• We train four different machine learning models using the cleaned text data:

  • Logistic Regression: A simple yet powerful model for binary classification.
  • Decision Tree: A tree-based model for decision making.
  • Random Forest: An ensemble method using multiple decision trees.
  • Gradient Boosting: A boosting algorithm to improve predictive accuracy.

• After training, we save the models and the vectorizer (for transforming new text input) using joblib.

• Users input news articles into the app. The text is preprocessed, transformed using the saved vectorizer, and passed through each of the models for classification.
• The app outputs whether the news article is True or Fake based on each model's prediction.

After training the models, we evaluate their performance using metrics such as:

• Accuracy: The proportion of correctly predicted news articles.
• Precision: The percentage of predicted fake articles that are actually fake.
• Recall: The percentage of actual fake articles that are correctly identified.
• F1 Score: A weighted average of precision and recall.

• Logistic Regression: 🏆 High Accuracy
• Decision Tree: 🌳 Balanced Results
• Random Forest: 🌲 High Precision
• Gradient Boosting: ⚡ Strong Recall

1. Input News Article: Type or paste any news article in the text box provided.
2. Click 'Analyze': Press the button to classify the article.
3. View Results: See predictions from each model. Whether the article is True ✅ or Fake ❌ will be displayed.

After setting up the environment and dependencies, you can test the application locally. Just run:

streamlit run fake_news_detection.py

This will launch the app in your browser.

The dataset consists of two parts:

• True News: Real articles (labeled as 1).
• Fake News: Misinformation articles (labeled as 0).

You can use this app to classify news articles based on their content.

We welcome contributions to improve this project! If you have suggestions, bug fixes, or new features to add, feel free to fork this repository and create a pull request.

This project is licensed under the MIT License - see the LICENSE file for details.

If you have any feedback or questions, feel free to reach out! You can contact the developer at (mailto:hardikarora483@gmail.com).