Fake News Detection Using Machine Learning

This project implements a comprehensive machine learning pipeline to classify news articles as true or fake. It utilizes traditional NLP techniques combined with multiple classifiers to detect misinformation effectively.

Features

• Text preprocessing: tokenization, lowercasing, punctuation removal
• Feature extraction using CountVectorizer and TfidfVectorizer with n-grams
• Classification models including:
  • Naive Bayes
  • Logistic Regression
  • Support Vector Machine (SVM)
  • Stochastic Gradient Descent (SGD)
  • Random Forest
  • Passive Aggressive Classifier
• Model evaluation with accuracy, F1-score, and confusion matrices
• K-Fold cross-validation for robustness
• Saving/loading models using pickle files
• Combining multiple models into a single pickle file for easy management
• Interactive command-line testing module with color-coded predictions
• Visualizations including bar charts and dynamic gauge charts (Truth-O-Meter)

Usage

1. Train models on the dataset or load pre-trained pickle files.
2. Use the interactive classify_news() function to input any news article and get predictions from all models.
3. Visualize results with helpful charts to understand prediction confidence.

Installation

• Python 3.x
• Libraries: scikit-learn, pandas, numpy, matplotlib, seaborn, plotly, nltk

pip install scikit-learn pandas numpy matplotlib seaborn plotly nltk