🧠 Sentiment and Emotion Analysis System

This project implements a hybrid Sentiment and Emotion Analysis System that combines traditional lexicon-based methods (TextBlob, VADER) with a transformer-based deep learning model (DistilRoBERTa from Hugging Face).

The system predicts both emotion categories (anger, fear, joy, love, sadness, surprise, neutral) and sentiment polarity (positive, negative, neutral) using a weighted ensemble strategy for improved accuracy.

---

🚀 Features

• Lexicon-based sentiment analysis using TextBlob and VADER
• Emotion classification using DistilRoBERTa (j-hartmann/emotion-english-distilroberta-base)
• Weighted ensemble method for combining model predictions
• Robust detection of neutral sentiment/emotion
• Evaluation using accuracy, precision, recall, F1-score, confusion matrix
• Clean, modular code with preprocessing, prediction, and evaluation

---

🧰 Technologies Used

• Python
• TextBlob
• VADER Sentiment
• Hugging Face Transformers
• scikit-learn
• matplotlib & seaborn
• Numpy
• Pandas
• Tensonflow
• Pytorch

---

🛠️ Setup & Usage

1. Clone the Repository

git clone https://github.com/saparya05/Employee-Management-System
cd sentiment-emotion-analysis

2. Activate Virtual Environment

# Linux/macOS
source env/bin/activate

# Windows
.\env\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

4. Run the Script

python main.py

📊 Sample Output

sample_texts = [I am so happy today! Everything is going perfectly."]

Analyzing: 'I am so happy today! Everything is going perfectly.'
Text: I am so happy today! Everything is going perfectly.
Predicted Emotion: joy
Confidence: 0.972

📁 Dataset

This project uses the Emotion Dataset from Hugging Face's datasets library. Contains 6 emotion labels: sadness, joy, love, anger, fear, and surprise Automatically loaded using:

from datasets import load_dataset
dataset = load_dataset("emotion")

📌 License

This project is open-source and available under the MIT License.