Model Deployment

Introduction

After training your machine learning model, the next step is to deploy it so others can use it in real-world applications. Deployment involves serving your model as an API so it can receive input data and return predictions.

Why Deploy a Model?

✅ Allow users and applications to use your model easily.
✅ Integrate machine learning into production systems.
✅ Automate predictions for workflows and products.

Tools for Deployment

Popular options include:

• FastAPI (lightweight, Python-based, easy to use)
• Flask
• Docker (for containerization)
• Cloud services (AWS, GCP, Azure)

In this tutorial, we’ll use FastAPI for simplicity.

Example: Deploying a Scikit-Learn Model with FastAPI

1️⃣ Install Dependencies

In your project environment, install FastAPI and Uvicorn:

pip install fastapi uvicorn scikit-learn pandas

2️⃣ Train and Save Your Model

First, train and save your model using scikit-learn:

# train_model.py
import pandas as pd
from sklearn.linear_model import LinearRegression
import joblib

data = {
    'Hours': [1, 2, 3, 4, 5],
    'Scores': [10, 20, 30, 40, 50]
}
df = pd.DataFrame(data)

X = df[['Hours']]
y = df['Scores']

model = LinearRegression()
model.fit(X, y)

# Save the model
joblib.dump(model, 'linear_model.pkl')

Run this script to save linear_model.pkl.

3️⃣ Create the FastAPI App

# app.py
from fastapi import FastAPI
from pydantic import BaseModel
import joblib

app = FastAPI()

# Load the trained model
model = joblib.load('linear_model.pkl')

# Request schema
class InputData(BaseModel):
    Hours: float

@app.post("/predict")
def predict(data: InputData):
    prediction = model.predict([[data.Hours]])
    return {"prediction": prediction[0]}

4️⃣ Run the FastAPI App

Run the server using Uvicorn:

uvicorn app:app --reload

Your API will be available at http://127.0.0.1:8000.

5️⃣ Test Your API

You can test your deployment using:

• The interactive docs at http://127.0.0.1:8000/docs
• Sending a POST request using curl or Postman:

curl -X POST "http://127.0.0.1:8000/predict" -H "Content-Type: application/json" -d '{"Hours": 6}'

Conclusion

🎉 You have successfully:

✅ Trained and saved a machine learning model.
✅ Created a FastAPI application to serve your model.
✅ Sent requests to get predictions from your deployed model.

What’s Next?

• Explore Docker to containerize your FastAPI app for easier deployment.
• Deploy to cloud platforms (AWS, GCP, Azure) for public access.
• Learn about model monitoring to track performance in production.

Join our SuperML Community to share your deployment projects and get feedback!