Beginner's guide to machine learning with Graphlab

Just like cloud computing ushered in the current explosion in startups … machine learning platforms will likely power the next generation of consumer and business tools.

We are living in a world where we are seeing vast advancements in ML and AI thanks to the democratization of machine learning, a form of artificial intelligence that enables computers to learn from data, without being explicitly programmed. With numerous online resources available, one often suffers from what right combination of tools to use to get started. This project aims to take away that friction and get you started with machine learning in minutes.

We will be using the following:

• Graphlab is a machine learning library from Turi which provides powerful tools for automatic feature engineering, model selection, and machine learning visualizations specific to the application and more.
• Docker is an awesome tool that you should have learnt yesterday. It makes setting up the development environment a breeze.
• Jupyter Notebook allows you to create and share documents that contain live code, equations, visualizations and explanatory text, right in your browser.

NOTE: If you want to use the open sourced Scikit-Learn with Pandas, refer to the ml-scikit-boilerplate.

To get started

• Install Docker
• Clone this project

git clone [email protected]:sud218/ml-graphlab-boilerplate.git
cd ml-graphlab-boilerplate

• Obtain your graphlab API key. It is completely free for personal use.
• Open .env file. Update GL_USER_EMAIL with the registered graphlab email id and GL_USER_KEY with the graphlab API key and save.

[email protected]
GL_USER_KEY=YOUR-GRAPHLAB-API-KEY

• Run make and you are done!

make

You should see the following with the link to your notebook. Grab that link and paste into your browser.

Voila! You are ready to get started on machine learning.

Note: If you are running on cloud, replace the ip 0.0.0.0 in the above url to with your docker-ip. You can very easily get your docker-ip by running make docker-ip.

What's next?

• When you open the notebook url, you will see some example notebooks. Go through them and run the notebooks to get first hand experience on different applications of machine learning.
• Read Graphlab docs and explore different models and tools available.
• Create your own notebook!

---

Facing trouble running? Please create an issue and I will get back to you.