13 April 2016
Coursera Machine Learning Specialisation: Part 1
Well, I have embarked on yet another series of courses looking at machine learning. After completing offerings from Johns Hopkins and Stanford University, I realise I may look a little keen. Further, the Stanford course was fairly in-depth, so why, might you ask, am I doing this?
A few reasons, and all pretty important to me:
- The course is taught emphasising the use of Python. I have gained significant experience using R over the last year or so, and to make myself more well rounded I want to get some practical experience using Python and relevant libraries for data science. After completing this course, I will find myself well exposed to Numpy, Matplotlib, and slightly less well known (but pretty powerful) libraries such as GraphLab Create and SFrame.
- The course revisits topics I wanted to explore in greater detail. For example, there is a six-week course dedicated to recommender systems and matrix factorisation, which was covered for only one week in Andrew Ng's course.
- The capstone project looks really neat, and it involves the development and deployment of a pretty cool recommender system. Where I have dabbled extensively with learning algorithms, I would say I am yet to really make the leap to having deployed a serious data product.
Course 1: Machine Learning Foundations: A Case Study Approach
Well, to date I have only completed the first course. My feelings about it are rather biased. To be honest, I probably have a background in machine learning that is more advanced than many who would have signed up to this course: a lot of this was revision to me. I may well have skipped this course if I didn't need it to unlock the capstone.
The instructors are pretty good, and the lectures are interesting. My only criticisms would be that the slides are at times a little 'too fun' for my taste, and that some of the introductory lecture videos could have been a little more polished- they come off more awkward than fun and spontanious. But seriously- do not let this put you off. The instructors are enthusiastic, and by combining the lectures slides with case studies using iPython notebooks they have created an engaging, informative course.
Also, I would like to point out that although many of the overall concepts are familiar to me, I am really looking forward to getting stuck in to the 'actual' modules. The will definitely go beyond my current level of knowledge, and I really think that I will benefit from this course, even having completed other machine learning courses prior to it!
Overall, a nice little introduction to the specialisation. I can see why this course was included- mainly to get everyone on to the same page before jumping into more advanced concepts. It may not have taught me loads of new material, but I enjoyed it, and I am really looking forward to completing the specialisation over the next few months.
TL;DR- this was a nice introductory course; I have high hopes for this specialisation! I will write another post in a few months to give my opinion once I have completed it.