It depends on your focus, of course. Andrew Ng's coursera is famous, and it's ideal for someone who wants to get into the mathematics behind various ML algorit

It depends on your focus, of course. Andrew Ng's coursera is famous, and it's ideal for someone who wants to get into the mathematics behind various ML algorithms. However, this class is will take you into implementing algorithms, but is less about applying them. If you want to just try them out, I'd honestly recommend just going through the scikit-learn documentation. Almost all of the algorithms provide an example, and the API is pretty consistent across different ML algorithms, to the extent that it can be. People learn differently, some people prefer to get into the math right away, others will never be interested in it. I'm interested, but I tend to be more motivated when I've used the algorithms, start to learn about how and why they perform well or poorly under various circumstances, and then dig into the mathematics specifically to find out why. Also, I'm not going to be creating new ML algorithms. So, you know, that also influences my level of interest. I do care about the mathematics involved, because I do want to genuinely understand why some outputs are available for random forests but not naive bases or logistic regression, why performance and/or accuracy is great in some circumstances and not others, and I don't want to have to rely on too much hand waving. But if you want to actually develop and research novel ML algorithms, you'd need to get considerably deeper into the math.