Welcome to an end-to-end machine learning dashboard app built for model evaluation and deployment — built for a hackathon, designed for production!

🔹 A powerful and stylish Streamlit dashboard for comparing multiple classification models.
🔹 Real-time model testing with uploaded CSVs.
🔹 Fully tuned pipelines, metrics analysis, and interactive visualizations.

The purpose of this project is to:

1. Train and evaluate multiple classification algorithms
2. Use cross-validation and hyperparameter tuning for optimization
3. Compare models based on metrics like:
   • Accuracy
   • AUC Score
   • F1-Score
   • Precision, Recall, Specificity
4. Visualize and interpret results through an interactive Streamlit dashboard
5. Enable end-users to upload their own CSV and get predictions from tuned models.

• ✅ Logistic Regression
• ✅ Decision Tree Classifier
• ✅ Random Forest Classifier
• ✅ Support Vector Machine
• ✅ XGBoost / LightGBM
• ✅ Hyperparameter Tuning (Grid Search)
• ✅ Feature Importance Charts
• ✅ Dynamic Bar Graphs (Plotly)
• ✅ Glassmorphic Streamlit UI
• ✅ Upload CSV to Test Models Live
• ✅ Auto-Pickle & Save All Models
• ✅ Responsive layout with dark mode and Fira Code font

Dataset is sourced from Kaggle. After selection:

• Null values handled
• Categorical features encoded
• Numeric features scaled
• Train/test split applied with stratification

Example input format is available in example_input.csv.

git clone https://github.com/yourusername/your-repo-name.git cd your-repo-name python -m venv venv source venv/bin/activate # or venv\Scripts\activate on Windows pip install -r requirements.txt

streamlit run app.py

Make sure models/ folder exists with pickled files

• Used Pipeline() from sklearn for each model
• Feature encoding, scaling, and classification in one step

• GridSearchCV for exhaustive tuning
• Best parameters auto-selected for each model

• Theme toggle: Light ✨ / Dark 🌚
• Plotly-based interactive charts
• Hover effects, rounded corners, and modern Fira Code font
• Upload .csv file to test any tuned model
• Performance chart comparison between Accuracy and AUC

Easily deploy on Streamlit Cloud:

https://share.streamlit.io/yourusername/your-repo-name/main/app.py

You can also deploy via:

• Hugging Face Spaces
• Render.com
• Local containerized environments (Docker)

1. Run app.py
2. Upload a CSV following the example_input.csv format
3. Select any model from the sidebar dropdown
4. Visualize predictions, performance, and insights

Dashboard View	Feature Importances

MIT License © 2025 Debangan Ghosh

Star ⭐ the repo if you liked the project. Contributions, feedback and forks are always welcome!

Connect with me on LinkedIn or drop an issue if you want to collaborate!