🐦 Twitter Sentiment Analysis (NLP Project)

This project performs Sentiment Analysis on Twitter data using Natural Language Processing (NLP) and a Naive Bayes Classifier to classify tweets into Positive and Negative sentiments.

---

📌 Project Overview

• Built a sentiment classifier for tweets
• Removed neutral tweets for better performance
• Applied text preprocessing and feature extraction
• Visualized frequent words using WordCloud

---

🧰 Tech Stack

• Python
• NumPy, Pandas
• NLTK (Natural Language Toolkit)
• Scikit-learn
• Matplotlib
• WordCloud

---

🔄 Workflow

1️⃣ Data Preprocessing

• Removed neutral tweets

• Cleaned text:

  • Removed URLs, mentions (@), hashtags (#)
  • Removed stopwords
  • Converted text to lowercase
  • Tokenization

---

2️⃣ Data Visualization

• Generated WordClouds for:

  • Positive tweets
  • Negative tweets

---

3️⃣ Feature Engineering

• Extracted words using nltk.FreqDist

• Created feature set using:

  contains(word)

---

4️⃣ Model Training

• Algorithm: Naive Bayes (NLTK)
• Trained on processed tweet dataset

---

5️⃣ Model Evaluation

📊 Results:

Training Size: 7449 tweets

Negative Tweets:
651 / 674 correct (~96.6%)

Positive Tweets:
51 / 176 correct (~28.9%)

---

⚠️ Observations

• High accuracy for Negative tweets

• Low accuracy for Positive tweets

• Possible reasons:

  • Class imbalance
  • Limited feature extraction
  • Simple model

---

🚀 Future Improvements

• Use advanced models:

  • Logistic Regression
  • SVM
  • Deep Learning (LSTM)

• Use TF-IDF / Word Embeddings

• Balance dataset

• Deploy using Streamlit

• Add real-time Twitter API

---

▶️ How to Run

git clone https://github.com/Mehtab161/Twitter-Sentiment-Analysis.git
cd Twitter-Sentiment-Analysis
pip install -r requirements.txt
jupyter notebook

---

📌 Requirements

See requirements.txt

---

👨‍💻 Author

Mehtab Khan

---

⭐ Support

If you like this project, give it a ⭐ on GitHub!