🐦 Twitter Sentiment Analysis using Python

📊 Project Overview

This project performs Sentiment Analysis on Twitter tweets to understand public opinions and emotional tone expressed in text data.
It involves data cleaning, EDA, text preprocessing, and training Machine Learning models — Logistic Regression and Random Forest — to classify tweets into Positive, Negative, or Neutral sentiments.

This was an independent solo project completed to strengthen my skills in Python-based data analysis, NLP, and supervised machine learning.

---

🎯 Objectives

• Clean and preprocess raw Twitter text data (remove stopwords, URLs, mentions, emojis, and punctuation)
• Perform Exploratory Data Analysis (EDA) to identify sentiment trends and frequent words
• Visualize insights using WordCloud and Seaborn visualizations
• Build and compare multiple ML models for sentiment prediction
• Evaluate models using accuracy, precision, recall, and F1-score metrics

---

🧠 Key Features

• Text Preprocessing: tokenization, lemmatization, stopword removal using NLTK & spaCy
• Visualization: WordCloud, sentiment distribution plots, and frequent word analysis
• Model Training: Logistic Regression & Random Forest Classifier
• Evaluation: Accuracy, Precision, Recall, and F1-Score comparison

---

⚙️ Machine Learning Model Performance

Model	Accuracy	Precision	Recall	F1-Score
Logistic Regression	0.7659	0.74	0.75	0.74
Random Forest	0.8850	0.87	0.88	0.87

✅ The Random Forest Classifier performed best, achieving 88.5% accuracy, showing strong capability to handle non-linear patterns in text sentiment.

---

🧰 Technologies & Libraries Used

• Programming Language: Python 3
• Libraries: Pandas, NumPy, Matplotlib, Seaborn, NLTK, spaCy, Scikit-learn, WordCloud
• Tools: Jupyter Notebook, GitHub

---

📈 EDA Highlights

• Visualized the distribution of sentiments across tweets
• Created a WordCloud to identify dominant words
• Explored frequent hashtags and mentions
• Identified correlation between text length and sentiment polarity

---

🗂️ Project Structure

twitter_sentiment_analysis/
│
├── twitter_sentiment_analysis.ipynb     # Main Jupyter Notebook with preprocessing, EDA, and ML models
└── data/
    └── tweets.csv                       # Raw dataset used for sentiment analysis

🚀 Future Improvements

• Experiment with advanced NLP models such as BERT or LSTM
• Enhance preprocessing to handle emojis, hashtags, and sarcasm detection
• Deploy the model as a web application for real-time tweet sentiment prediction
• Add model explainability (SHAP/LIME) to interpret sentiment decisions
• Include hyperparameter tuning for improved accuracy and generalization

🧑‍💻 Author

Deepak Kumar
📧 Email : [deepak.kumar8434543@gmail.com]
🔗 LinkedIn : [www.linkedin.com/in/deepak-kumar-acb2002]
🌐 GitHub : [https://github.com/Deepakkumar165]