How much math is used in machine learning

If you’re looking to get into machine learning then you might be wondering how much mathematics will be involved.

This post will show you how math heavy machine learning is, how much you will need to know and how you can go about learning it and getting started in machine learning if you don’t.

So, how much math is used in machine learning? To understand the machine learning algorithms it will be necessary to have an understanding of calculus, linear algebra, probability and statistics.

The amount of mathematics that you will need to know will be largely impacted by what you intend to do in machine learning. If you want to be able to apply the machine learning models to data then it will just be necessary for you to have an undergraduate level understanding of the above. If you want to be a machine learning researcher then it will be necessary for you to have a graduate-level understanding of statistics as well.

datacamp banner

How much math you need to know for machine learning

If you want to be able to apply machine learning models to do things such as compete in machine learning competitions or to even get an entry-level job then the amount of mathematics that you will need to know will lower.

In this case, it will not be necessary for you to learn each of the subjects in full detail and there will only be subsections of the courses that it will be necessary for you to know.

Linear algebra

Out of all of the mathematics subjects that you need to know for machine learning, I would say that linear algebra is the most important one since it gets used constantly. In linear algebra it will be necessary for you to understand vector spaces, linear transformations, multiplying matrices, the dot product, how to get the inverse of a matrix, eigenvalues and solutions to linear systems.

The video below will take you through all of the linear algebra that you need to know for machine learning.

Calculus

In calculus, it will be necessary for you to learn about limits, differentiation, integrals, and partial derivatives.

You can watch the video below, on youtube, that will take you through all of the calculus that you need to know for machine learning.

Probability and Statistics

It will be necessary for you to know most of the topics that get taught in probability theory and statistics at an introductory level. This means that it would help to have an understanding of probability distributions, PDFs, CDFs, independence, joint probability, central limit theorem and the law of large numbers.

The best course that I would recommend on probability would be this course on EDX that is taught by MIT. It will be necessary for you to learn calculus before taking the course so make sure to learn calculus first.

How to get started in machine learning

There are actually courses, in machine learning, that do not assume any prior knowledge from you.

One such course is an introduction to machine learning taught be Andrew Ng.

This is the most popular machine learning course that there is and I would highly recommend that you start by taking this course before you even start learning the mathematics detailed above.

This is because you will be able to learn a lot about the different machine learning algorithms and whether or not machine learning is for you. By doing this you will be able to avoid spending a lot of time learning the required mathematics before realizing that machine learning actually isn’t for you.

If you then decide that machine learning is something that interests you then it would be worthwhile for you to take the time to learn the mathematics in more detail. Once you have learned the mathematics in more detail then I would recommend that you take this machine learning course which is taught by MIT. It goes into the mathematics of machine learning in more detail and makes use of a lot of probability theory.

After having taken Andrew Ng’s course it would be worthwhile for you to start learning how to apply the machine learning algorithms in a popular programming language such as python. A very good book that I would recommend that you read on how to apply machine learning algorithms, in Python, would be this one. It has the prerequisites of knowing, calculus, linear algebra and Python so you will need to learn those first.

Once you have learned how to apply the machine learning algorithms in Python then it would be worthwhile for you to start applying your skills on real datasets. One way to do this would be to compete in machine learning competitions on the website Kaggle.

So to summarize the steps that I would recommend that you take are:

  1. Take Andrew Ng’s machine learning course
  2. Learn calculus, linear algebra, probability theory and Python
  3. Work through either Hands-on Machine learning or take MIT’s machine learning course (or do both)
  4. Apply what you know on Kaggle

What other skills do I need for machine learning?

The other skills that you will need to know will be to program in a popular machine learning programming language such as R or Python and how to do some data analysis in that programming language.

I would recommend that you learn Python if you want to learn machine learning as it is the most popular machine learning language with a lot of support online. The course that I would recommend for learning Python would be this one which is taught by MIT on EDX.

To learn the data analytics that you need to know for machine learning then I would recommend that you watch this Youtube series.

Related

What math is needed for data science?

Understanding of calculus, linear algebra, probability and statistics, at an undergraduate level, is what would be necessary for data science. It will also be very important for you to learn data analysis and to be proficient in a programming language such as Python.

How long will it take me to learn machine learning?

It takes, on average, 55 hours to complete Andrew Ng’s machine learning course so that is all the time that is necessary to learn what the machine learning algorithms are and how they work.

However, to apply machine learning algorithms it will be necessary for you to also know data analysis and a programming language. So, you will also need to spend a similar amount of time learning these if you don’t know them already.

If you want to get a good mathematical understanding of the machine learning algorithms and the statistical techniques involved in figuring out how to apply them then it will also be necessary for you to learn linear algebra, calculus, probability and statistics. You should expect these courses to take at least 55 hours each to learn as well if your goal is to just get an overview of how to make use of them.

It would also be worthwhile to take the time to work through a course on how to apply the algorithms in a programming language such as with MIT’s course or Hands-on Machine Learning. MIT’s course lasts for 15 weeks and requires 10 to 14 hours of effort per week.