Books on machine learning

machinelearning
keras

#1

We’ve had some great threads on books for learning statistics and other topics recently! Machine learning is obviously a massive trend right now, and packages like keras look super exciting.

The trouble is, I have zero background in this stuff, and seeing some minimal examples of a package running isn’t really the same as understanding what machine learning is trying to achieve and how it’s going about it (perhaps contrasted with, say, regression modelling, which I feel I understand pretty well). I’m looking for some summer reading material (since, y’know, it’s not like I have a thesis to submit or anything :cry:) to take me a bit further than the ML elevator pitch. Does anyone have any favourite books? I’m fairly language-agnostic (R would be preferred with Python a runner up, but again, I’m more interested in how well the theory is laid out). Cheers!


What are the goto resources for Machine learning in R
#2

A great starting place that I’d recommend would be An Introduction To Statistical Learning; it’s a very well-written and accessible introduction to the area, complete with R exercises and code for you to get more familiar with. If you want a heavier version of the material (with more of the maths left in) from the same authors, check out The Elements of Statistical Learning.

For a nice, practical, online course introducing the basic concepts I’d recommend Machine Learning on Coursera; its fantastic content with some excellent hands-on exercises, and will give you some of the finer details and broader concepts, too.


#3

edX also has a few worth checking out. This one from caltech is basically filmed lectures but the guy knows his stuff. Others have good reviews but I haven’t tested them personally

I’d definitely be interested to know if anybody has a bookdown project underway featuring relevant R packages


#4

I second “An Introduction To Statistical Learning” as a good start. The authors of the text teach an free online course that follows the material of the book. I found this a very engaging course which gave me a good introduction to the subject. I wasn’t expecting it, but it also contains a lot of important supplementary material on statistical methods such as bootstrapping, cross validation etc which are important for the field, but not immediately thought of when putting together a curriculum on machine learning.

n.b The course covers methods such as linear regression, decision trees, and support vector machines. It doesn’t really touch on “Deep learning” such as neural nets.


#5

I can highly reccomend the book from Max Kuhn (the author of the caret pkg) - Applied Predictive Modeling. Apart from showing and discussing the models it also goes into the practicalities of training, pitfalls to watch out etc.


#6

This is probably the best resource I found for Neural Networks/Deep Learning: http://neuralnetworksanddeeplearning.com/chap1.html

I find it well-written and easy to dive into, compared with say, the Goodfellow book on Deep Learning.


#7

An Intro to and Elements of Statistical Learning came up in an earlier thread, so I reckon they’ll be a good start :slight_smile: I also like the look of that link, @pedram! Thanks, everyone :smiley:


#8

I 100% support starting with ISLR. It’s a great guide to the fundamentals and underlying theory.

For book 2, I recommend:

It’s a great walk through a dozen ML models serving a specific purpose. It skips over some theory in favor of getting you to be productive. It also really doesn’t require much programming experience.

For book 3, I recommend going to ESL. It’s much denser than the other two books, but covers the theory in great detail. If you get through those three books, you’ll have a solid understanding of the underlying statistics and the practical usage of machine learning in R. From there, picking up new tools and techniques, like neural nets using Keras, will make a lot more sense and be easier.


#9

If you are interested in deep learning and keras, then the upcoming book “Deep Learning with R” is an excellent choice.


#10

Some people here have linked to specific O’Reilly books, but I can’t see a link to their free ebooks page, which is fantastic. It covers ebooks from “O’Reilly editors, authors, and Strata speakers” covering topics such as Data Science, AI and Big Data.


#11

A post was split to a new topic: Questions about "Introduction to Statistical Learning" and about Machine Learning Generally