https://github.com/iakshatgandhi/fake-news-classification-model-main
A machine learning-based project designed to classify news articles as real or fake. This system combines advanced natural language processing (NLP), robust machine learning models, and intuitive visualizations to deliver accurate and scalable predictions.
https://github.com/iakshatgandhi/fake-news-classification-model-main
matplotlib nltk pickle python scikit-learn seaborn
Last synced: 24 days ago
JSON representation
A machine learning-based project designed to classify news articles as real or fake. This system combines advanced natural language processing (NLP), robust machine learning models, and intuitive visualizations to deliver accurate and scalable predictions.
- Host: GitHub
- URL: https://github.com/iakshatgandhi/fake-news-classification-model-main
- Owner: iakshatgandhi
- License: apache-2.0
- Created: 2024-12-30T18:26:57.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2025-01-17T08:55:19.000Z (10 months ago)
- Last Synced: 2025-10-09T18:33:10.955Z (24 days ago)
- Topics: matplotlib, nltk, pickle, python, scikit-learn, seaborn
- Language: Python
- Homepage:
- Size: 244 KB
- Stars: 1
- Watchers: 1
- Forks: 1
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Fake News Classification Project
==========================
**Project Overview**
This project utilizes natural language processing (NLP) and machine learning algorithms to classify news articles as real or fake. It features a user-friendly web interface built with Streamlit for effortless interaction.
**Features**
* **Data Preprocessing:** Thorough cleaning and preprocessing of data to prepare it for modeling.
* **Text Vectorization:** Utilizes TF-IDF (Term Frequency-Inverse Document Frequency) to convert text data into numerical features.
* **Classification Models:** Explores multiple models including Naive Bayes, Logistic Regression, and Random Forest for optimal classification performance.
* **Model Evaluation:** Compares and analyzes the performance of different models using metrics like accuracy, precision, recall, and F1-score.
* **Visualizations:** Generates informative visualizations to understand data distribution, model performance, and feature importance (for Random Forest).
* **Web Interface:** Streamlit-based interface allows for real-time classification of news articles directly through a web browser.
**Installation**
1. **Clone this repository.**
2. **Install required packages:**
```bash
pip install -r requirements.txt
```
3. **Download NLTK data:**
```python
import nltk
nltk.download('punkt')
nltk.download('stopwords')
nltk.download('wordnet')
```
**Usage**
1. **Prepare dataset:** Place files named `True.csv` and `Fake.csv` containing your real and fake news data in the project directory.
2. **Run data preprocessing and model training script:**
```bash
python data_cleaning.py
```
3. **Launch Streamlit app:**
```bash
streamlit run app.py
```
4. **Open web browser:** Navigate to the provided URL (usually `http://localhost:8501`) to access the web interface.
**File Descriptions**
* `data_cleaning.py`: This script handles data preprocessing, model training, and evaluation.
* `app.py`: This script builds the Streamlit web application for user interaction.
* `fake_news_model.pkl`: Stores the best-performing trained model for real-time classification.
* `tfidf_vectorizer.pkl`: Stores the fitted TF-IDF vectorizer used for text feature extraction.
**Model Performance**
The project evaluates the performance of multiple classification models:
* Naive Bayes
* Logistic Regression
* Random Forest
It analyzes the models using various metrics:
* Accuracy
* Precision
* Recall
* F1-score
**Visualizations**
The project generates insightful visualizations to understand the data and model behavior:
* Model performance comparison
* Dataset distribution (real vs. fake news)
* Text length distribution
* Learning curves
* Feature importance (for Random Forest)
These visualizations are accessible through the Streamlit web interface.
**Web Interface**
The Streamlit web interface allows users to:
* Input news articles for real-time classification.
* View the classification result (real or fake) along with a confidence score.
* Explore model performance metrics and visualizations.
**Future Improvements**
* Implement cross-validation for robust model evaluation.
* Experiment with deep learning models like LSTM and BERT.
* Add features like sentiment analysis and named entity recognition.
* Enhance the web interface with interactive elements.
* Implement user feedback collection for continuous improvement.
**Contributing**
We welcome contributions! Feel free to submit a Pull Request to enhance this project.
**License**
This project is licensed under the Apache License 2.0 - see the LICENSE file for details.