I've been going through the course and realized that many of the exercises are lacking depth, so I intend on re-writing most of the algorithms without Octave/Matlab so I can get a genuine understanding of the process.
Current I'm doing a lot of web development with Ruby on Rails, but I'd like to break into machine learning. I have a strong feeling that mastering (not just doing the exercises, but practicing them regularly) the algorithms that Andrew Ng has shown in the course might be enough that I may end up with one or two opportunities to use these skills while doing my job, say for example classifying users based on site activity or being able to predict the amount of traffic on any given day..
However, I'm being lured into a lot more other challenging courses, like coursera's NLP course. I don't want to end up a jack-of-all-trades with little chance of actually applying what I've learned to my daily job, so I feel focusing on a small number of algorithms to master would be better than getting a general understanding of many algorithms.
Has anyone else gone the web development to machine learning route? How as it worked out for you? How did you manage to break in?
TLDR: Do I have to learn lots of algorithms to become an effective ML-er or know a handful really well?
[link] [22 comments]