Machine Learning Flask Deployment on Microsoft Azure 2021

Learn to prep machine learning models using Python and scikit-learn with a diabetes dataset, handling missing values and skewness, and building a production-ready pipeline for deployment on Microsoft Azure.

Analyze dataset with exploratory data analysis using correlation matrices and Seaborn visuals. Examine distributions, outliers, skewness, and median or mean handling across glucose, BMI, insulin, age, pregnancy, and diabetes outcomes.

Upsample the minority diabetes class to balance the dataset, split into train and test with stratification, train multiple classifiers, evaluate with accuracy and confusion matrices, and save models for deployment.

Test saved models by loading them from local files, performing predictions, and evaluating accuracy with confusion metrics. Build a production-ready pipeline by generating deployment-ready code from this workflow.

Build and test a flask app in VS Code that loads a saved diabetes model with pickle, feeds eight features (sex, bmi, pregnancies, age, etc.), and outputs predictions.

Create production-ready python class with a predict method for diabetes-pedigree-function, accepting glucose, blood pressure, skin thickness, insulin, bmi, and age, loading model for flask deployment on azure.

Create a basic Flask web app by configuring a folder, naming the app apple pie, installing Flask, and setting up templates and routes with get and post methods.

Design a front-end web interface for a Flask-based machine learning app, building a form with dropdowns and input fields, wiring fields to model selection and feed into a backend pipeline.

Apply CSS styling to your web frontend by creating a stylesheet, wiring it into templates, and using flex, gradients, borders, and responsive form design to enhance inputs, labels, and buttons.

Apply media base configurations to CSS using media queries to ensure responsive, mobile-friendly layouts across web and mobile browsers by integrating base styles into the project structure.

Relocate the CSS file for a Flask server by creating a static folder and placing CSS and JavaScript assets there, then verify paths to ensure the front-end loads correctly.

Develop a Flask backend to receive diabetes prediction inputs via post, integrate a prediction function, and return results on the local server, laying groundwork for html display and deployment.

Transform a Flask backend into a web front end by returning JSON messages from a data list and rendering results, prepped for Azure deployment.

Create a free Azure account, configure a Python 3.8 web app with a new resource group, and enable GitHub continuous deployment for a Flask deployment.