If you’ve worked all the way through this course, you should now consider yourself an expert in Machine Learning

Probably not gonna make that claim yet, but the journey has been an interesting one. To recap…

1. Signed up with Coursera on June 23rd but did nothing with it.
2. Enrolled in the critically acclaimed Machine Language course by Andrew Ng.
3. Digging up fragmented memories of linear algebra and calculus which fortunately have only been tucked away and not conveniently forgotten.
4. First real Linear Regression assignment on Week 2 - completed Aug 8th.
5. Got slammed hard by Neural Networks assignment on Week 5 - completed Aug 30th.
6. Sped through the rest of the content and finishing the course 2.5 weeks later.

And the results:

So, What Have I Learnt?

Supervised Learning - linear regression, logistic regression, neural networks, support vector machines

Unsupervised Learning - k-means, principal component analysis, anomaly detection

Special Applications - recommender system, large scale machine learning (map-reduce / on-line learning)

Building a ML System - bias/ variances, regularization, evaluate learning algorithms, learning curves, error analysis, ceiling analysis

More Importantly, What’s Next?

This course has provided a good firm foundation on the concepts and ideas of Machine Learning and Predictive Analytics. However these are also, IMHO, very academic and the tools used are quite rough.

As Predictive Analytics and other types of advanced analytics technology are becoming major factors in the analytics market, we have big vendors providing enterprise grade advanced analytics platforms that are more approachable for non propeller heads.

Having said that, my next step would still be leaning towards venturing into R Programming. Because… 10 reasons why you should learn R.

Notes / Other Sources

Machine learning notes from holehouse.org

comments powered by Disqus