#120 on Reddit
Statistical Learning
Stanford Lagunita
15 recommenders · 14 threads · recommended in 5 years, 2015–2019 · score 17.6
An introductory Stanford online course by Trevor Hastie and Rob Tibshirani covering core statistical learning methods, taught practically in R, with a free companion textbook (An Introduction to Statistical Learning). Commenters describe it as a statistics-oriented complement to more computation-focused machine learning courses.
Commenters call it a light, practical, free intro that explains methods from a statistics perspective and pairs well with the ISL book.
Recommendations by year
20152019
It will be job dependent, but in a true DS role, math - specifically statistics is a fundamental building block. If I had the choice to hire someone to be a DS based simply on degree - and I had a CS major and a Stats major, I would take the stats major every time. Start by reading through this…
It depends. When you are training with dropout, your training data changes all the time, hence, it will fluctuate, but the general trend should be go down, then, go up. If no dropout, and you do not shuffle your dataset, the curve will be smooth with the same trend, but the result probably won't…
They cover similar topics, but not always the same. Ng goes into neural networks which are very powerful. This may be an unpopular opinion but I believe that the statlearning course is the better option here. but, it definitely feels like it has more of a statistics feel to it than Ng's course. IF…
Take the Stanford course by the authors. The book is free and you'll learn a lot. https://lagunita.stanford.edu/courses/HumanitiesScience/StatLearning/Winter2014/about
ISLR has coding examples and exercises (in R). The authors also offer a Statistical Learning MOOC: https://lagunita.stanford.edu/courses/HumanitiesSciences/StatLearning/Winter2016/info. If you take the StatLearning class, you'll learn much more than taking a scikit-learn class.
That's a great course! You should definitely try to complete the exercises to get the most of it. Most of the content is downloadable as well, so you should have everything you need as you go. Also, a very good complementary course is Stanford's *Statistical Learning* by Trevor Hastie and Rob…
I'm in the same position as you; I'm interested in Data Science, but I need some guidance. I've done a bunch of searching on the topic, and I've found the two of the most important topics in math are Linear Algebra and Statistics. For Linear Algebra, I recommend MIT OCW's 18.06: Linear Algebra…
I'd recommend getting Introduction to Statistical Learning (it's free! http://www-bcf.usc.edu/~gareth/ISL/ ) while taking the stat learning course on Stanford Lagunita. https://lagunita.stanford.edu/courses/HumanitiesSciences/StatLearning/Winter2016/about If you're looking for a more mathematical…
Do a project. The textbook that actually gets you to the finish line on that project is the best one. Do another project after that. Don't procrastinate waiting and searching for "the best" of anything. You might just be putting things off unnecessarily. By the way you should add Johns Hopkins…
I highly recommend to start with this one: https://lagunita.stanford.edu/courses/HumanitiesandScience/StatLearning It gives a lot of simple but yet interesting insight. Then try Andrew NG course on Coursera from Standford. First, get confidence in making things happen.
On a side note, if you are particularly interested in statistical learning, I would suggest ISL and the expanded ESL by Hastie and Tibshirani might be a more worthwhile option than Andrew Ng's course (https://lagunita.stanford.edu/courses/HumanitiesSciences/StatLearning/Winter2016/about). It also…
Look up some basic R tutorials and with some luck you'll find some trends in your data. I'll warn you, it requires knowledge of Stastics to take full advantage of R. If you're willing to take the plunge I recommend you go through this lecture series by Stanford on Stastical Learning There's a free…
It depends on what you want to be doing, where you want to be doing it and how well you want to be doing it. In general: the more math you know, the better. * If you just want to use some simple methods / do basic data science analysis, you won't be needing much math (maybe high-school level…
These two courses are VERY different even though they cover some of the same tools. Andrew Ng's [Coursera] Machine Learning course is excellent at giving you a high level understanding of what various machine learning methods actually do along with some fundamental concepts like Bias/Variance…
Hi, I was kind of in your shoes a year ago, when I decided to embark on a Machine Learning journey through MOOCs and online resources. Here's my opinion: * Stanford Statistical Learning - light and practical course, you will learn how to do apply the algorithms using R with two amazing professors…
I've taken the former, not the latter. I'd say that the former starts you thinking about some cool data science-y things (basic visualization and simulation stuff), but doesn't go into that much depth. I thought it was worth taking but it won't teach you nearly as much about ML as e.g. Andrew Ng's…
I am now three years down the road that you are planning to follow and I have landed my first ML job this year. So here is my personal experience and my suggestions that follows from it. I also started with DL courses on fastai, udacity and udemy. They are all pretty good; they get you familiar…
I have some views on this.. will write to you in a bit here.. short answer is : depends on your background and what do you intend to do with it. Brb Edit: My background: 10+ years in trading at a bank. Undergrad commerce + and MBA in Finance. But very interested in automation for a while. Learnt…