You need machine learning then you’re going to need to learn maths, people often ask me what’s the best programming language for doing machine learning and obviously I think Python is a great choice but you know choice of language is really far less important than understanding the maths behind machine learning.

Machine learning is all about maths, and in this article I want to tell you what maths is involved, why you need it and how to learn it. To really understand machine learning you’re going to be confident in linear algebra, calculus, probability, and statistics.

Why linear algebra ?

Well computers interact with the world very differently from the way that we do, where we see an image, a computer sees a two or three dimensional matrix, so virtually every bit of input that goes into a computer is stored as a matrix and linear algebra concerns the manipulation of matrices and that’s something that computers have become very good and they’re very fast at it, and so that’s why you need to learn linear algebra, you have this in out that comes in that needs some kind of processing, because when you take a dataset whether it’s from an image whether it’s a sound whatever you’re working with all that is a collection of data, it’s eat data and you have to do something with data.

So you might want to do some sort of feature construction on that data, so you can create features from that data that are more useful to you, in order to do that you might have to transform, combine, or break down that data in some way to create new data that’s hybrid of the data that you started with.

And also you’ve got this set of data that you really now insight in it again, it’s just a load of raw numbers . so you’re need to perform some kind of descriptive statistics on that data, so you understand the data a little bit better so you might visualize that, you need to plot histogram of that data, you might want to just mine that data a little bit more for some information to find out where to go with the next stage of machine learning process, and so you can build a predictive model so that you can predict from unseen data what likely outcome is going to be.

So e need to look into this data and use descriptive statistics to do that, but it doesn’t stop there, we might need to change the data a little bit by normalizing it, that invites statistics too, we might need to use something called hot encoding and then you have understand the math behind the actual algorithms themselves, you know why might you learn linear regression, what are the strengths of linear regression when would you use logistic regression.

**Next article i’m gonna to talk about the best ressources to learn ML.**