Integrating ML Models

Integrating ML Models#

This document explains how to integrate a machine learning model into a FastAPI application. It includes a simple example to make it easy to understand. The goal is to create an API that receives input data, processes it through a pre-trained machine learning model, and returns predictions.

👉 New to App-Generator? Sign IN with GitHub or Generate Web Apps in no time (free service).

Scenario: Let’s consider a simple scenario where we want to predict whether a person has diabetes based on some medical features. We’ll use a pre-trained machine learning model (for simplicity, we will use scikit-learn).

Steps:

  1. Train a simple machine learning model.

  2. Save the trained model to a file.

  3. Create a FastAPI app to serve predictions from the model.

Prerequisites#

  • Python 3.7+

  • FastAPI

  • Uvicorn (for running FastAPI)

  • Scikit-learn

  • Joblib (for saving and loading the model)

Install dependencies:#

pip install fastapi[all] scikit-learn joblib uvicorn matplotlib

Step 1: Train a Simple Model#

We will start by training a simple machine learning model to predict whether a person has diabetes based on their medical features.

# train_model.py

import joblib
import matplotlib.pyplot as plt
from sklearn.datasets import load_diabetes
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier

# Load diabetes dataset
diabetes_data = load_diabetes()

# Visualize the distribution of the target variable
plt.hist(diabetes_data.target, bins=50)
plt.xlabel('Target Value')
plt.ylabel('Frequency')
plt.title('Distribution of Target Variable')
plt.show()

# Convert target into binary (1 = diabetic, 0 = not diabetic)
y = diabetes_data.target > 100

# Split into training and testing sets
X = diabetes_data.data
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# Train a RandomForestClassifier
model = RandomForestClassifier()
model.fit(X_train, y_train)

# Save the trained model to a file
joblib.dump(model, 'diabetes_model.joblib')

print("Model trained and saved.")

Run this script to train and save the model:

python train_model.py

Step 2: Create FastAPI Application#

Now, we will create a FastAPI application that loads the saved model and exposes an endpoint to make predictions.

# main.py

import joblib
import numpy as np
from fastapi import FastAPI
from pydantic import BaseModel

# Load the saved model
model = joblib.load('diabetes_model.joblib')

# Create FastAPI app
app = FastAPI()


# Define the input data structure
class DiabetesFeatures(BaseModel):
    age: float
    sex: float
    bmi: float
    map: float
    tc: float
    ldl: float
    hdl: float
    tch: float
    ltg: float
    glu: float


# Define a prediction endpoint
@app.post("/predict/")
def predict(features: DiabetesFeatures):
    # Convert input data to a numpy array
    input_data = np.array([[  # Convert input features to the appropriate format
        features.age,
        features.sex,
        features.bmi,
        features.map,
        features.tc,
        features.ldl,
        features.hdl,
        features.tch,
        features.ltg,
        features.glu
    ]])

    # Get the prediction probability (probability for "diabetic")
    prediction_prob = model.predict_proba(input_data)

    # Set a threshold for predicting 'Diabetic' (e.g., 0.6 probability for being diabetic)
    threshold = 0.6
    is_diabetic = prediction_prob[0][1] > threshold  # Use the second column for the probability of class '1'

    # Return the result based on probability threshold
    return {"prediction": "Diabetic" if is_diabetic else "Not Diabetic"}

Step 3: Run the FastAPI Server#

To start the FastAPI server, use uvicorn:

uvicorn main:app --reload

Step 4: Test the API#

Once the server is running, you can test the API using curl or through an interactive UI at http://127.0.0.1:8000/docs.

Example curl request:

curl -X 'POST' \
  'http://127.0.0.1:8000/predict/' \
  -H 'Content-Type: application/json' \
  -d '{
  "age": 50.0,
  "sex": 1.0,
  "bmi": 25.0,
  "map": 92.0,
  "tc": 220.0,
  "ldl": 120.0,
  "hdl": 50.0,
  "tch": 70.0,
  "ltg": 3.5,
  "glu": 90.0
}'

Expected Response:

{
  "prediction": "Not Diabetic"
}

Conclusion#

In this guide demonstrates how to integrate a machine learning model with FastAPI. The key steps were:

  1. Train and save a machine learning model.

  2. Load the model in a FastAPI application.

  3. Expose an endpoint to make predictions based on input data.