The Coursera MOOC platform offers two specializations in the field of Data Science, that I have considered for learning about Data Science. One is the Johns Hopkins University Data Science Specialization, and the second is the University of Illinois at Urbana-Champaign’s Data Mining Certificate.
The two specializations take different approaches, the Johns Hopkins’ attempting to cover more ground while the Urbana-Champaign’s is more specific.
Both specializations are composed of courses that can be joined free of charge, however certification has a (minimal) cost. Joining the specialization is one options, otherwise all the courses can be taken individually.
The first difference between joining the specializations and taking all courses independently is the possibility to participate to capstone projects after completion of all courses of the specialization. The second is the possibility of a certificate on the specialization, which may be a valid asset in many cases. I suggest to whoever is serious in doing this to take the Specialization approach and go all the way, however do not try to do too many things at once because the courses need some work and it is hard to conciliate a daily job, often travelling around the globe with studying in the evening. So, do not try to do too much at the same time and stay focused (Antonio’s advice). What is my approach then? Am I doing what I say? Well, for the moment I am not taking the specialization or the certified learning. As I mentioned before I have a daily job and I do this just because I like it. But I also discovered the world of MOOCs not so long ago, so I was not sure about my capacity to keep the pace.
So, then what are the differences between the two specializations?
The Urbana-Champaign Data Mining Certification is composed of:
- Pattern Discovery in Data Mining
- Text Retrieval and Search Engines
- Cluster Analysis in Data Mining
- Text Mining and Analytics
- Data Visualization
- And finally the Data Mining Capstone
These courses require a certain amount of mathematical background, and go quite deep. For the two courses that I have enrolled in (Text Retrieval and Search Engines and Text Mining and Analytics) the programming assignments are performed using a C++ framework called the META toolkit, and you need to work in C++, which may be a show stopper for many. However, the programming part is not mandatory, but I think that to get a better understanding of the subjects the best way is to get your hands dirty and the programming assignments are the best way to do it. I have also enrolled in Pattern Discovery in Data Mining but I did it too late and I could not complete it, so I am waiting for another chance.
So, in brief, this specialization go straight at the heart of Data Science, of which Data Mining is one of the most interesting aspects and gives you the tools to comprehend most of the problems at hand. This is one perspective.
The Johns Hopkins’ Data Science Specialization is composed of:
- The Data Scientist’s Toolbox
- R Programming
- Getting and Cleaning Data
- Exploratory Data Analysis
- Reproducible Research
- Statistical Inference
- Regression Models
- Practical Machine Learning
- Developing Data Products
- And finally the Data Science Capstone
The approach is from the ground up, the first course gives an initial bag and is in my opinion a prerequisite for most of the others. The R programming language is one of the most popular data analysis and visualization languages on the web (I do not know it yet) and there are many flames arguing which one is better between this one and python with its scientific libraries. From what I have read around, my feeling is that R is a tool that a Data Scientist has to have in his toolbox. I have started learning python because I found a good set of resources but I intend to take this specialization, I do not know yet if as single courses or as a whole. Committing to take and pass 9 courses is not an easy undertaking, but the advantage of this one over the Urbana-Champaign (which I hope are being re-offered but I do not see any dates yet) is that the JH courses are almost monthly, so the accessibility is extremely high. So, from this point of view the JH gives some advantages. I also believe that the subjects treated are giving a bigger amplitude to the specialization, after all Data Science is bigger than Data Mining, I have not followed yet any of these (but I have chatted with friends who are doing it), so I will come back on this later.
