This project is a simple yet powerful implementation of a linear regression model served through a FastAPI application. It's designed to predict outcomes based on a pre-trained TensorFlow model. The API is intuitive and documented, making it accessible for developers and machine learning enthusiasts.

• FastAPI Framework: Utilizes FastAPI for efficient and easy-to-use API endpoints.
• TensorFlow Integration: Leverages a TensorFlow model for making predictions.
• Auto-Generated Documentation: Includes Swagger UI documentation generated automatically by FastAPI.

Before you begin, ensure you have met the following requirements:

• Python 3.8 or above
• pip for package installation

1. Clone the repository to your local machine:

   git clone https://github.com/hiroyukikumazawa/linear-predict.git

2. Navigate to the project directory:

   cd linear-predict

3. Install the required packages:

   pip install -r requirements.txt

To start the FastAPI server, run the following command:

uvicorn app:app --reload

This command will start the development server with live reloading. The API documentation will be available at http://127.0.0.1:8000/docs.

To make predictions, use the /predict/ endpoint. You can do this through the Swagger UI or by sending a POST request with a JSON body containing the values for prediction:

{
  "values": [0.1, 0.5, 0.9]
}

If you wish to retrain the model with your data, run:

python ./train/train.py

This script generates synthetic training data, trains the linear regression model, and saves it for the application to use.

We welcome contributions! If you have suggestions for improvements or bug fixes, please open an issue or submit a pull request.

Distributed under the MIT License. See LICENSE for more information.

Hiroyuki Kumazawa - [email protected]

Project Link: https://github.com/hiroyukikumazawa/linear-predict

• FastAPI
• TensorFlow