Getting Started With Hands-On Machine Learning

I am often asked about resources that would be helpful for someone who is new to ML/AI. Obviously there are tons of resources available online but that is exactly why it can be overwhelming for those who are just getting started. Here’s my curated list of resources for anyone with some technical background but no ML/AI experience. Note that there are two areas areas: First is the analysis — sort of a traditional statistician or data-heavy analyst work — and the second is more software focused i.e. building machine learning models. Nothing stops people from first area to become expert on second area. The focus of this post is more about machine learning. The key thing is to get started and learn by doing!

Prerequisites:

  1. Linear Algebra (College level)
  2. Basic Statistics and Probability (College level)
  3. Elementary knowledge of SDLC, some programming experience and familiarity with cloud
  4. Elementary Data analysis, manipulation and database knowledge

Now that we have covered these, lets get to the building blocks. I will not cover details here but still wanted to mention that you need to remember that ML/AI work is part of a bigger puzzle.

  • Looking at the big picture – business need, application of results
  • AI work and how it overlaps with other disciplines and areas
  • Modeling Life cycle and Work Management

Key Technical Goals

  • Understand technical stack
  • What personal / cloud setup is needed to get started?
  • Options for learning on your own – keep reading for more!

Top Free Resources for ML and AI

To keep things simple I’ll skip resources that go deep into the theory of machine learning models. Its important to learn the pros and cons of different models and when to use A vs B or a combination of models. At the same time beware of spending too much time open-ended discussions, edge cases and finer details. I recommend 80-20 rule. To keep things simple, I’ll only talk about programming in Python. Here are 6 recommended tasks for you.

  • Use Google Colab if you want to skip installing software package on your computer for local work. If you are comfortable with installing software development packages and command line then I highly recommend installing Conda. Why? it comes with the libraries and packages that you will need to create models.
  • Take ML Crash Course by Google
  • Run sample code using Seedbank: tools.google.com/seedbank
  • Take Andrew Ng’s ML (and more) courses on Coursera – it offers both free and paid option.
  • Try MS Azure ML
  • Fast.AI is a good resource if you have 1+ year of python experience

This is just the beginning. I kept this post short on purpose so that you can focus on learning the fundamentals without getting distracted. If you are done with the above or feel that you need more advanced sources then look for upcoming posts.

Scroll to top