Seems like the kind of stuff you get at by combining sklearn's official docs with some material from one of the moocs (like udacity's machine learning class, coursera's etc.) + some wikipedia. This is my default style of learning: picking a "track" (whatever mooc or a cool side project idea) + lots of small specific resources.
I'm really curious why would you recommend and actual book?
It has all the chances to get outdated in a few months, while the separate resources will get updated for sure. yeah, there's some googling effort, but still...
Most (all? I can't remember exactly) of it doesn't use Scikit-learn. You actually implement the algorithms yourself and see what happens under the hood
I just got this book ~2 weeks ago and am using it in conjunction with udacity's course. Python Machine Learning (PML) is more low-level, math intensive, and hands on than udacity and other resources I'm using.
Thus far Udacity's course fits your explanation of using sklearn docs, tweaking inputs, etc. while PML provides a technical explanation of weighting with perceptrons and then you hand-write a perceptron learning algorithm.
Books are able to dig into a single topic deeply in a connected way with a single author's voice who can assume at any point all the given knowledge from the previous chapters. It is possible to learn from many shallow or specific sources, but it's not as easy/much more effort for most.
They give away one book per day. They don't send an email, but I have a script that scrape their site and send myself an email. I'm happy to add to a mailing list if there's interest.
Just a thought.. instead of building something that sends out an email... it would be much more convenient to build something that automatically adds each new book to your account.. using something like Python's Selenium library to do it.
> Is this part of a Christmas season special, or do they regularly do giveaways?
I specifically remember them doing something like "20 books in 20 days" at the end of the last year, where you could get one book per day.
Seems like that run was quite successful and that they are doing it regularly now. I have like 21 books in my account without actually buying any of them.
EDIT: Correction, I've just checked my order history and it turns out that giveaway I was referring to was at the end of 2014, not 2015.
Funny, Just discovered and bought it yesterday :-) Seems like a very good book just by glancing at contents as well as the way the material is presented and taught. Noticed the Amazon reviews are also very good.
I don't know anything about this book, but in general the pre-reqs for any form of ML (unless you are working in the abstract or with a "black-box" library or such) are a good working knowledge in statistics/probabilities, and an understanding of linear algebra (mainly how to manipulate and work with vectors/matrices).
Oh - and an understanding of when and how to "parallelize" problems can be important, too.
ML and associated problems aren't the most intuitive to understand, solve, etc - but they are a great challenge to expose yourself to if you have the interest.
/disclosure: I'm currently enrolled in and participating in Udacity's Self-Driving Car Engineer nanodegree, plus I have participated in and completed their CS373 program, and I was a part of (and completed) the original Stanford ML Class MOOC that was taught by Andrew Ng. I guess you can say I am a bit biased on this subject...?
I would recommend knowledge in Python. If you don't know Python prior to reading this book, then you won't learn it here.
The Linear Algebra presented in the book is easy, but if you aren't familiar with Linear algebra then it would be a benefit to learn some basics before reading.
Familiarity with mathematical notation and mathematical reasoning is also a must. If you don't have a basic understanding of being able to follow mathematical arguments, then some side comments (that aren't explicitly stated) may seem mysterious to you.
This book is very suitable for someone with a background in Python and basic Linear Algebra/mathematical reasoning knowledge.
The point of this book is to understand the math/algorithms and not treat the algorithms as blackbox solutions. You'll learn about processing your data, dimension reductions, etc., etc.
You may have to do some outside studying depending on your background, but the author provides those resources for you in the text. Overall this is a very good book and the author did a good job at writing it.
I recently bought this book because I found Andrew Ng's course to be a bit too theoretical and mathematical. I quickly found out that this book is VERY math heavy as well though.
Does anyone have any resources to get into ML without having to understand a lot of maths? I just want to understand the concepts, the pros/cons of the different algorithms, and how to use them with some common libraries/services that exist today (TensorFlow?).
26 comments
[ 1.8 ms ] story [ 65.5 ms ] threadIs this part of a Christmas season special, or do they regularly do giveaways? And do they send emails about the free book of the day?
I'm really curious why would you recommend and actual book?
It has all the chances to get outdated in a few months, while the separate resources will get updated for sure. yeah, there's some googling effort, but still...
Thus far Udacity's course fits your explanation of using sklearn docs, tweaking inputs, etc. while PML provides a technical explanation of weighting with perceptrons and then you hand-write a perceptron learning algorithm.
[1]: https://news.ycombinator.com/item?id=13107226
[1]: https://www.changedetection.com/
I specifically remember them doing something like "20 books in 20 days" at the end of the last year, where you could get one book per day.
Seems like that run was quite successful and that they are doing it regularly now. I have like 21 books in my account without actually buying any of them.
EDIT: Correction, I've just checked my order history and it turns out that giveaway I was referring to was at the end of 2014, not 2015.
https://news.ycombinator.com/item?id=6933716
https://datahack.analyticsvidhya.com/contest/ama-with-sebast...
Oh - and an understanding of when and how to "parallelize" problems can be important, too.
ML and associated problems aren't the most intuitive to understand, solve, etc - but they are a great challenge to expose yourself to if you have the interest.
/disclosure: I'm currently enrolled in and participating in Udacity's Self-Driving Car Engineer nanodegree, plus I have participated in and completed their CS373 program, and I was a part of (and completed) the original Stanford ML Class MOOC that was taught by Andrew Ng. I guess you can say I am a bit biased on this subject...?
The Linear Algebra presented in the book is easy, but if you aren't familiar with Linear algebra then it would be a benefit to learn some basics before reading.
Familiarity with mathematical notation and mathematical reasoning is also a must. If you don't have a basic understanding of being able to follow mathematical arguments, then some side comments (that aren't explicitly stated) may seem mysterious to you.
This book is very suitable for someone with a background in Python and basic Linear Algebra/mathematical reasoning knowledge.
The point of this book is to understand the math/algorithms and not treat the algorithms as blackbox solutions. You'll learn about processing your data, dimension reductions, etc., etc.
You may have to do some outside studying depending on your background, but the author provides those resources for you in the text. Overall this is a very good book and the author did a good job at writing it.
Does anyone have any resources to get into ML without having to understand a lot of maths? I just want to understand the concepts, the pros/cons of the different algorithms, and how to use them with some common libraries/services that exist today (TensorFlow?).