This post is a bit behind schedule but better late than never as they always say. On October 15th I completed the Machine Learning by Stanford University course on Coursera. I had been interested in making a start in machine learning and hoped that the course would provide a good foundation for building upon.
The course is taught by Andrew Ng, an adjunct professor at Stanford University and former Chief Scientist at Baidu. It aims to be a “broad introduction to machine learning, datamining, and statistical pattern recognition”. It largely achieves that with 11 weeks of material spanning gradient descent to neural networks. Most weeks consist of a combination of video lectures, readings, guided programming assignments, and quizzes. An interesting feature of the video lectures are inline mini-quizzes that engage the learner and prevent them from zoning out when being introduced to particularly challenging material. Programming assignments are completed in Octave and for each assignment the course provides tutorials and unit tests. There is also an active forum should the learner require tailored assistance.
I found that the course approached the topic of machine learning in a logical, well-organized manner. The learner is introduced to the “why” of machine learning and given a 10,000 foot view of the field of machine learning. Andrew Ng then, over the course of 11 weeks, drills down a few 1000 feet into various topics in the field. Having completed the course I am no expert in the field, but feel that I have firm foundation for future growth. The teaching was so good in fact, that I was able to successfully submit a kernel to Kaggle’s “Titanic: Machine Learning from Disaster” introductory competition! This was a feat I could not complete in my tries before the course and I achieved it solely through applying the knowledge and methods instilled through the course.
If you are interested in a start in machine learning and can dedicate yourself to 10-12 hours per week of study you will be rewarded. My journey took longer than the projected 11 weeks (life happens) but Coursera provides the option for continuing your learning in another session. I can’t recommend the course enough and feel it should be the first step for any aspiring machine learner.