Documentation API

Introduction

This API aims to expose a machine learning model. It contains a single endpoint allowing a prediction to be made from the explanatory variables. It includes token-based authentication.

Code Structure

• main.py: The entry point of the API application. It initializes the FastAPI application, sets up routing, and may include logic for starting the server.
• predict.py: Contains the logic for making predictions with the machine learning model. It likely defines FastAPI endpoints to receive prediction requests and return the results.
• opentelemetry_setup.py: Configures OpenTelemetry for monitoring and tracing the application. It helps in collecting metrics and traces for monitoring and debugging.
• utils.py: Contains utility functions used across the API application. This may include authentication logic, helpers for data processing, etc.
• database.py: contains the logic for connecting to and interacting with the database. It may include functions for creating database sessions, data models, and CRUD operations
• model_loader.py: contains the logic for loading the machine learning model from a file or external source. It's used to prepare the model for predictions.
• launch_app.sh: A shell script used to launch the API application. It checks if the machine learning model is loaded and, if not, runs a script to load it before starting the API with Uvicorn.

Authentification

You need a token to use the predict endpoint of the api. To get this token you need to call the generate_token("admin") function in the api/utils.py file

python3 -m api.utils

Endpoints

Prediction

• URL: /predict
• Method: POST
• Authentication required: Yes
• Request body: Model-specific data for prediction.
• Response: Prediction result.