🌟 Twitter Sentiment Analyzer — Explainable AI Dashboard

Advanced NLP pipeline using TF-IDF + Logistic Regression with Explainable AI token attribution, batch CSV processing, and a live Streamlit dashboard.

---

🔗 Live Demo

👉 Try it live →

Type any text or upload a CSV and get instant sentiment predictions with explainability!

---

🎯 Overview

A fully interactive Sentiment Analysis Dashboard that goes beyond basic positive/negative prediction — it shows which words drove the sentiment using Explainable AI token attribution.

Built for:

• Social media analysis (tweets, reviews)
• Customer feedback classification
• Batch processing of large text datasets

---

✨ Features

🔍 Real-Time Text Analysis

• Enter any text and get instant prediction
• Positive / Negative classification
• Confidence score with visual gauge bar

📊 CSV Batch Processing (Report Studio)

• Upload any CSV with a text column
• Full predictions table with sentiment labels
• Sentiment distribution bar chart
• Positive + Negative word clouds
• Dataset summary metrics

🧠 Explainable AI (Token Attribution)

• Highlights top positive words boosting sentiment
• Highlights top negative words lowering sentiment
• Weighted importance score per word
• Makes the model transparent and interpretable

🎨 Modern Dark-Themed UI

• Professional dashboard layout
• Two-tab interface: Live Analyzer + CSV Studio
• Clean typography and animated components

---

🏗️ System Architecture

Raw Text Input
      │
      ▼
Text Cleaning (NLTK)
      │
      ▼
TF-IDF Vectorization
      │
      ▼
Logistic Regression Model
      │
      ├──► Sentiment Prediction (Positive/Negative)
      ├──► Confidence Score
      └──► Token Weight Attribution (Explainability)
                  │
                  ▼
         Streamlit Dashboard

---

⚙️ Tech Stack

Component	Technology
ML Model	Logistic Regression (scikit-learn)
Feature Engineering	TF-IDF Vectorizer
Explainability	Token Weight Attribution
NLP Preprocessing	NLTK
Visualizations	Matplotlib, WordCloud
UI & Deployment	Streamlit + Streamlit Cloud

---

📁 Project Structure

Sentiment-Analysis/
│
├── app.py                     ← Streamlit dashboard (main entry)
├── sentiment_utils.py         ← Prediction + Explainability engine
├── train_model.py             ← Model training script
├── test_analysis.py           ← Local testing script
├── requirements.txt
│
└── models/
    ├── sentiment_model.joblib     ← Trained model
    └── tfidf_vectorizer.joblib    ← Fitted vectorizer

---

🚀 Run Locally

# Clone the repo
git clone https://github.com/M1325-source/Sentiment-Analysis.git
cd Sentiment-Analysis

# Install dependencies
pip install -r requirements.txt

# Run the app
streamlit run app.py

Open http://localhost:8501 in your browser.

---

📊 Model Performance

Metric	Score
Algorithm	Logistic Regression
Features	TF-IDF (n-grams)
Dataset	Twitter Sentiment140
Accuracy	~85%+

---

🔮 Future Enhancements

• BERT / Transformer-based model upgrade
• Multi-class sentiment (Very Positive, Neutral, Very Negative)
• Real-time Twitter API integration
• Aspect-based sentiment analysis
• REST API endpoint for programmatic access

---

👩‍💻 Author

Manisha Priya — AI/ML Engineer

• 🐙 GitHub: @M1325-source
• 📧 Email: manishapriya1325@gmail.com

⭐ Star this repo if you found it useful!