How to learn AI and machine learning

Learning AI and machine learning means learning how models learn patterns from examples, built on a base of Python, NumPy and pandas, and some algebra, linear algebra and statistics. Google's Machine Learning Crash Course lists these foundations:

A robot head in profile made of glowing circuit-board lines
A robot head drawn in circuit patterns, a picture of the systems that learn from data.Photo: DeltaWorks / Pixabay
  1. 1

    Python

    With NumPy and pandas.

  2. 2

    Algebra

    Variables, equations, functions.

  3. 3

    Linear algebra

    Vectors and matrices.

  4. 4

    Statistics

    Mean, median, standard deviation.

Learn in Evernote. Clip the docs, let the AI Assistant fix your code, and search any snippet.

Start your coding notebook in Evernote

How to learn AI and machine learning in 6 steps

  1. 1

    Learn Python first

    Google's Crash Course recommends some programming experience in Python, because its exercises are in Python. Learn the basics, then NumPy and pandas for working with data.

  2. 2

    Brush up the math

    Google's Machine Learning Crash Course lists algebra, linear algebra and statistics as prerequisites, with calculus optional.

  3. 3

    Start with the concepts

    If you are new, Google suggests its short Introduction to Machine Learning before the Crash Course.

  4. 4

    Train a first model

    A linear regression on a small table, like car weight and miles per gallon. Learn what features, labels and fitting mean.

  5. 5

    Use browser notebooks

    The Crash Course runs its exercises in Colaboratory, in the browser, with no setup.

  6. 6

    Build a small project

    Pick a dataset you care about, train a simple model and write down what worked and what did not.

A 6-week plan to start machine learning

About 45 minutes a day, following Google's prework
  1. 1Weeks 1 to 2Python basics, then the NumPy and pandas tutorials.
  2. 2Week 3Review algebra, vectors and basic statistics.
  3. 4Week 4First model: linear regression on car weight and miles per gallon.
  4. 5Weeks 5 to 6Machine Learning Crash Course modules, with the Colab exercises.

Free coding practice log

One entry per session. Copy it, or save it to your Evernote account and keep your snippets in the same note.

Date: [day] · Time spent: [minutes]

Topic: [one concept, like for loops or functions]

Source: [tutorial, chapter or video, with its link]

What I built: [a small program, in one line]

Error I hit: [the last line of the error message]

How I fixed it: [one sentence, plus the corrected snippet]

Next session: [the next concept to learn]

Trusted by Millions Worldwide

4.4

2,100+ reviews on G2

4.4

8,200+ reviews on Capterra

4.4

73,000+ reviews on App Store

248M

Registered users

AI learning questions

Can I learn AI for free?

You can start with Google's Machine Learning Crash Course in the browser, and Harvard's CS50 courses are free through OpenCourseWare.

What do I need to know before learning machine learning?

Python with NumPy and pandas, plus algebra, linear algebra and statistics. Google lists calculus as optional.

What is the difference between AI and machine learning?

Machine learning is a part of AI in which systems learn patterns from data instead of following hand-written rules.

Do I need a powerful computer?

Not to start. Google's Crash Course runs its exercises in Colaboratory, in a web browser.

What is a model in machine learning?

A program that has learned a relationship from examples, such as car weight to miles per gallon, and uses it to make predictions.

How long does it take to learn machine learning?

Each Crash Course module is self-contained, and Google lists an estimated module length of 80 minutes for linear regression alone. How long the whole course takes depends on your schedule.

How to learn AI and machine learning with Evernote

Keep one notebook for machine learning: the course pages you clip, your practice code, and every error you fixed.

  1. 1

    Clip the regression lesson.

    Save the linear regression lesson with the Web Clipper. Keep it in a notebook called Learning machine learning, with the car weight example and its link.

  2. 2

    Ask why fit() fails.

    Put the model code and its TypeError in a note. Ask the AI Assistant why fit needs a label, and accept the added y argument.

  3. 3

    Find fit again.

    Search for "fit" when you need the method again. Evernote finds it in your Google clip and in the practice note, code blocks included.

  • You can also ask the AI Assistant to quiz you on a chapter you clipped, one question at a time.
  • You can also log your projects in a table, with a link to each one and what you learned.

Every snippet, error and fix in one place

Clip the docs you learn from, ask the AI Assistant why your code breaks, and find any snippet again by searching for it.

Start for free