First of all we do not need to write the complete code for each algorithm, there are libraries that do that part quite effectively and hopefully are easy to mess around with, I use python ( still need to start with scikit ) but i find it rather easy to mess around in it. Even though we don't need to implement the algorithm we should know the math behind it so that we can judge which algo will be more efficient for which data set and also set the desired perimeters to improve the result (eg alpha in the first exercise).
The most important part of a job would be collecting the data .i.e. data mining, which is not covered in this course. Though is mining much different from data scraping? I use python to scrape some data from a few sites now and then, which is kind of fun for me.
Knowing which part of the data do you need to compute the desired results .i.e. the features, the rite features can help get the result faster and more efficiently. I find the idea of Kernels very intriguing but I still don't know much about them or how I would use kernel equations in a real challenge.
[link][3 comments]