Iris ML API

FastAPI service that exposes a machine learning model for Iris flower classification using scikit-learn.

The goal of this project is to demonstrate how I:

• train a machine learning model
• serve it through a REST API
• automatically test the API
• integrate the project with continuous integration

---

🚀 Tech Stack

Tool	Purpose
Python 3.x	Programming language
FastAPI	Web framework for serving the API
scikit-learn	Machine learning library
joblib	Model serialization
pytest	Automated testing framework
GitHub Actions	Continuous Integration

---

📁 Project Structure

ml-iris-api/

├── app/
│   ├── __init__.py
│   └── main.py          # FastAPI app loading the trained model and exposing /predict
│
├── models/
│   └── iris_rf.joblib   # trained RandomForest model
│
├── tests/
│   ├── __init__.py
│   └── test_api.py      # API tests using TestClient
│
├── train_model.py       # script to train and save the model
│
├── requirements.txt
├── pytest.ini
│
└── .github/
    └── workflows/
        └── tests.yml    # CI pipeline

Directory Overview

app/ Contains the FastAPI application and prediction endpoint.

models/ Stores the trained machine learning model.

tests/ Automated API tests using pytest and FastAPI TestClient.

train_model.py Script used to train the machine learning model and save it.

GitHub Actions workflow Continuous Integration pipeline configuration.

---

🤖 Training the Model

To train the machine learning model locally:

pip install -r requirements.txt
python train_model.py

This script:

1. Downloads the Iris dataset
2. Trains a RandomForest classifier
3. Saves the trained model to:

models/iris_rf.joblib

---

⚡ Running the API Locally

Start the FastAPI server:

uvicorn app.main:app --reload

The API will run at:

http://127.0.0.1:8000

---

📚 API Endpoints

Healthcheck

GET /health

Returns a simple response confirming that the service is running.

---

Prediction Endpoint

POST /predict

Example request:

{
  "sepal_length": 5.1,
  "sepal_width": 3.5,
  "petal_length": 1.4,
  "petal_width": 0.2
}

The API returns:

• predicted Iris class
• class probabilities

Possible classes:

• setosa
• versicolor
• virginica

---

📊 Interactive API Documentation

FastAPI automatically generates interactive documentation.

Swagger UI

http://127.0.0.1:8000/docs

OpenAPI schema

http://127.0.0.1:8000/openapi.json

These allow testing the API directly from the browser.

---

🧪 Tests

The test suite verifies key functionality of the API.

Covered scenarios:

• /health endpoint
• valid prediction request
• invalid input validation
• API response correctness

Run all tests:

pytest

Tests use FastAPI TestClient, which allows testing the API without starting a separate server.

---

🔁 Continuous Integration

The repository includes a GitHub Actions CI pipeline.

Workflow location:

.github/workflows/tests.yml

Pipeline steps

Triggered on:

• push
• pull_request to the main branch

Pipeline actions:

1. Install project dependencies
2. Train the ML model using train_model.py
3. Run the full pytest test suite

This ensures that:

• the model trains correctly
• the API works
• the tests pass on every commit

You can view pipeline runs in the Actions tab of this repository.

---

🛠 Future Improvements

Potential improvements for the project:

• add Docker containerization
• add model versioning
• integrate ML experiment tracking (MLflow)
• add input schema validation
• deploy the API to a cloud environment

---

👨‍💻 Author

Kacper Blok

Machine Learning / Backend / QA Automation Portfolio Project