README.md

metadata

title: SVM Regressor
emoji: 🤖
colorFrom: purple
colorTo: blue
sdk: streamlit
sdk_version: 1.41.1
app_file: app.py
pinned: false
license: mit
short_description: App demonstrating the SVM Regressor

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

SVM Regressor Demo 🤖

This app demonstrates the use of Support Vector Machines (SVM) for regression problems in various domains. It provides a user-friendly interface to select a problem, visualize the dataset, train an SVM Regressor, and interpret the results.

Features

• Problem Selection: Choose from three example problems: Business 💼, Engineering ⚙️, or Education 🎓.
• Dataset Display: View the corresponding dataset for the selected problem, along with a description of its features.
• Model Training and Performance: Train an SVM Regressor on the selected dataset. Display the model's performance using metrics like R-squared and Mean Squared Error. Adjust hyperparameters to fine-tune the model. Visualize the regression results using a scatter plot with a regression line.
• Output Interpretation: Get a clear explanation of the model's output in the context of the selected problem. Understand the significance of the results and how they can be used for predictions or decisions.

Learning Objectives

• Understand the concept of SVM Regression and its applications.
• Learn how to train and evaluate SVM Regressors using Python.
• Explore the impact of different hyperparameters on the model's performance.
• Interpret and explain the results of SVM Regression in various domains.

Datasets

The app uses three synthetic datasets for demonstration purposes:

• Business 💼:
  • Features: Usage (in hours), Demographics (binary)
  • Target: Churn (continuous)
  • Description: Predicts customer churn based on usage patterns and demographics.
• Engineering ⚙️:
  • Features: Sensor_Data (continuous)
  • Target: RUL (Remaining Useful Life) (continuous)
  • Description: Predicts the remaining useful life of an industrial machine based on sensor data.
• Education 🎓:
  • Features: Study_Hours (continuous), Prev_Grades (continuous)
  • Target: Test_Scores (continuous)
  • Description: Predicts student performance on a standardized test based on study habits and previous grades.

How to Use

1. Select a Problem: Choose one of the three problems from the dropdown menu.
2. View Dataset: The corresponding dataset will be displayed, along with a description of its features.
3. Train and Evaluate Model: Adjust the hyperparameters (kernel, C, epsilon) using the sliders. The model will be trained on the dataset and its performance metrics will be displayed.
4. Visualize Results: The regression results will be visualized using a scatter plot with a regression line.
5. Interpret Output: Read the interpretation of the model's output in the context of the selected problem.

Additional Features

• Interactive Visualizations: Explore the impact of different hyperparameters on the model's performance by adjusting the sliders and observing the changes in the regression results.
• Upload Your Own Data: Train an SVM Regressor on your own dataset by uploading a CSV file. (This feature is not yet implemented.)

Requirements

• Python 3.7+
• Streamlit
• Pandas
• Scikit-learn
• Matplotlib

Installation

1. Install Python 3.7+
2. Create a virtual environment (optional but recommended)
3. Activate the virtual environment
4. Install the required packages using pip: pip install streamlit pandas scikit-learn matplotlib

Usage

1. Download the repository and unzip it.
2. Navigate to the repository directory.
3. Run the app using streamlit run app.py.

Contributing

Contributions are welcome! If you find any bugs, have suggestions for improvements, or want to add new features, please feel free to submit a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

Acknowledgements

This app is inspired by https://medium.com/coinmonks/support-vector-regression-or-svr-8eb3acf6d0ff.