Sentiment Analysis on Twitter Tweets (Sentiment140 Dataset)

🚀 Project Overview

This project focuses on Sentiment Analysis of Tweets using the popular Sentiment140 dataset.
The model predicts whether a tweet expresses a positive or negative sentiment by leveraging Natural Language Processing (NLP) techniques and a Logistic Regression classifier.

The pipeline includes:

• Text preprocessing (cleaning, stopword removal, and stemming)
• TF-IDF Vectorization for numerical feature extraction
• Model training using Logistic Regression
• Model evaluation on unseen test data
• Model persistence with pickle for future use

---

📂 Dataset

• Dataset: Sentiment140
• Size: 1.6 million tweets
• Target Variable:
  • 0 → Negative Sentiment
  • 4 → Positive Sentiment (converted to 1 in this project)

---

🛠️ Tech Stack

• Languages: Python 3.10
• Libraries:
  • numpy, pandas – Data handling
  • nltk – Stopwords, stemming
  • scikit-learn – TF-IDF, train-test split, Logistic Regression
  • pickle – Model saving
• Environment: Google Colab

---

⚙️ Project Pipeline

1. Data Loading: Load the dataset with correct encoding (latin-1).
2. Data Cleaning:
   • Remove unwanted characters, mentions, URLs, and punctuation.
   • Apply stemming using PorterStemmer.
3. Feature Extraction: Convert text to numerical vectors using TF-IDF.
4. Train-Test Split: 80% training, 20% testing (stratified).
5. Model Training: Logistic Regression with max_iter=1000.
6. Evaluation:
   • Training Accuracy: ~80%
   • Test Accuracy: ~77%
   • No significant overfitting detected.
7. Model Deployment: Save model as trained_model.sav for re-use.

---

📊 Model Performance

Dataset	Accuracy
Training	80.4%
Testing	77.7%

---

🔍 Example Predictions

# Load the model
import pickle
loaded_model = pickle.load(open('trained_model.sav', 'rb'))

# Predict sentiment
tweet = "I love this product! Absolutely amazing."
vectorized_tweet = vector.transform([tweet])
prediction = loaded_model.predict(vectorized_tweet)

print("Positive Tweet" if prediction[0] == 1 else "Negative Tweet")