A.I. Sommelier

Visit App on Heroku

PROJECT CASE

Can A.I. accurately predict red wine quality rating?
Anyway, what is quality if not how wine taster's tongue precepts wine's chemical components?

Let's try to answer that question with machine learning approach, where various classifier algorithms will try to discover all the patters in wine rating process. As a last part of a project, let's build some artificial sommeliers and let them handle real-world wine samples. Everything in a form of python flask application.

APP STRUCTURE

Flask App has 4 views, each of them representing different stage of machine learning process. Each section ends with downloadable code template.

Part 1 - Dataset Overview

First, we'll have a closer look on dataset. With numpy, pandas, seaborn and scikit-learn:

Part 2 - Building Classifier

In the next step we'll build foundations of our classifier while choosing best fitting algorithms:

Part 3 - Fitting Classifier

Now, it's a time to test our classifiers and see how they're performing:

Part 4 - Rating Red Wine

Let me introduce you to our A.I. Sommeliers:

TOOLS, MODULES & TECHNIQUES

Python – web development:

flask | Conda | Heroku | Docker

Python – DB:

flask_sqlalchemy

Python – data analysis & visualisation:

pandas | numpy | matplotlib | seaborn

Python – machine learning:

sklearn | scipy | pandas | numpy | pickle

Database Development:

SQLite

Testing

selenium | unittest

Web Development:

HTML | CSS | Bootstrap | Materialize | JavaScript | JQuery

Test Suite:

Travis CI:

Test Files:

/tests

Thank you,

Lukasz Malucha