📰 Fake News Detection using Logistic Regression

A binary text classification machine learning project that detects whether a news article is real or fake using Natural Language Processing (NLP) techniques and Logistic Regression.

---

🧠 Problem Statement

Given a dataset of real and fake news articles, build a model that can classify new/unseen news as either real or fake based on its textual content.

---

📊 Dataset Details

• Source: WELFake_Dataset.csv
• Download Link: WELFake Dataset on Kaggle
• Columns Used: title, text, label
• Target Label:
  • 1 = Real news
  • 0 = Fake news
• Final Input Feature: Combined title and text into a single text field before preprocessing

---

🚀 Project Workflow

1. Import Dependencies

2. Load Dataset (WELFake_Dataset.csv)

3. Preprocess Data

   • Remove nulls
   • Combine text columns
   • Clean text with regex
   • Remove stopwords & apply stemming

4. Feature Engineering

   • TF-IDF Vectorization using unigrams & bigrams

5. Train-Test Split

   • Stratified 80/20 split

6. Model Training

   • Compared Logistic Regression, Naive Bayes, and Random Forest
   • Selected Logistic Regression for best balance of speed and performance

7. Model Evaluation

   • Accuracy, Precision, Recall, F1-score
   • Confusion Matrix (visualized with seaborn)

8. Save Model & Vectorizer using pickle

9. Custom Prediction

   • In-notebook prediction function

---

📈 Model Performance

Metric	Training Set	Test Set
Accuracy	95.86%	94.64%
F1-score	0.96	0.95

🔍 Confusion Matrix Heatmap (Test Set)

✔️ Indicates strong generalization with balanced performance on both classes.

---

🧪 Example Usage

Enter: "Breaking: President gives major update on national policy."
Output: "Prediction for custom news input: Real "

---

📦 Installation & Setup

git clone https://github.com/Toshaksha/fake_news_prediction.git
cd fake_news_prediction
pip install -r requirements.txt

---

🗂 Project Structure

fake-news-detection/
│
├── fake_news_prediction.ipynb            # Jupyter notebook (model training)
├── requirements.txt                      # Python dependencies
├── models/
│   ├── logistic_regression_model.pkl     # Saved ML model
│   └── tfidf_vectorizer.pkl              # Saved TF-IDF vectorizer
├── images/
│   └── confusion_matrix.jpg              # Confusion matrix heatmap
└── README.md                             # Project documentation

---

🧰 Tools & Libraries Used

• Python 3.x
• NLTK – for stopwords and stemming
• Scikit-learn – ML models and metrics
• Pandas, NumPy – data handling
• Seaborn, Matplotlib – visualization
• tqdm – progress bar for processing

---

👤 Author

Toshaksha – GitHub Profile

---

⭐ If you found this project helpful, please give it a star on GitHub!