📰 Fake News Prediction using Machine Learning

This project builds a machine learning model to classify news articles as real or fake using Natural Language Processing (NLP) and Logistic Regression.

---

🔍 Problem Statement

Fake news poses a significant challenge in the digital world. This project aims to classify news articles based on their text content into two categories:

• Real News (label = 0)

• Fake News (label = 1)

⚠️ Note: The model's predictions are limited to the provided dataset and may not generalize well to real-world news articles.

---

📁 Dataset

• Source: Kaggle Fake News Dataset

• File used: train.csv

• Features: id, title, author, text, label

---

🛠️ Technologies Used

• Python (Google Colab)

• Numpy, Pandas

• nltk for stopwords removal and stemming

• scikit-learn for TF-IDF, model training, and evaluation

---

⚙️ Data Preprocessing

1. Missing Values: Replaced with empty strings

2. Content Creation: Combined author and title into a new content feature

3. Text Cleaning:

• Remove non-alphabetic characters

• Convert to lowercase

• Tokenize

• Remove stopwords

• Apply stemming

4. Feature Extraction: Used TfidfVectorizer to convert text into numerical form

---

🤖 Model Details

• Algorithm: Logistic Regression

• Data Split: 80% training, 20% testing

• Input: TF-IDF features from preprocessed text

• Evaluation: Accuracy score

---

📊 Results

• Successfully trained and evaluated the model.

• Achieved good accuracy on the test set from the dataset.

---

🧪 Usage

• Upload train.csv to your Colab session

• Run the notebook cells sequentially

• The notebook handles all preprocessing, training, and prediction

---

🚧 Limitations

• The model is trained and tested only on the provided dataset.

• It may not accurately classify news from outside sources due to:

  • Dataset bias

  • Lack of contextual understanding

  • No real-world generalization capabilities

---

🔮 Future Improvements

• Experiment with advanced models (SVM, XGBoost)

• Use deep learning models (LSTM, BERT)

• Apply cross-validation and hyperparameter tuning

• Train on more diverse and recent data sources

---