Twitter sentiment analysis using all-MiniLM-L6-v2 sentence embeddings. This project compares multiple machine learning models for sentiment classification (positive, negative, neutral) and includes comprehensive visualizations.
- Text Preprocessing - Complete tweet cleaning pipeline
- Sentence Embeddings - Using all-MiniLM-L6-v2 (384-dim vectors)
- 4 Classification Models - Logistic Regression, XGBoost, SVM, Random Forest
- Cosine Similarity Analysis - Embedding space visualization
- 2D Visualization - UMAP dimensionality reduction
- Custom Predictions - Test with your own tweets
| Model | Accuracy | F1-Score |
|---|---|---|
| SVM (RBF) | 68.20% | 0.684 |
| Logistic Regression | 66.40% | 0.663 |
| XGBoost | 63.50% | 0.637 |
| Random Forest | 61.80% | 0.618 |
- Python 3.x
- sentence-transformers
- scikit-learn
- XGBoost
- UMAP
- matplotlib, seaborn
All contributions are welcome — bug fixes, feature enhancements, or documentation improvements!
Please give appropriate credit to the original author if you use or modify this tool in your own projects.
This project is licensed under the MIT License — see the LICENSE file for details.
Ayush Mani
🔗 GitHub: @m1n1v1rus
# Clone repository
git clone https://github.com/m1n1v1rus/Sentiment-Embeddings-Project.git
# Install requirements
pip install -r requirements.txt
# Run notebook
jupyter notebook Sentiment_Embeddings_Project.ipynb