The Data Scientist’s Toolbox
Getting and Cleaning Data
Exploratory Data Analysis
Part of the Data Science Specialization
Provider: Coursera (John Hopkins University)
Tutors: Roger D. Peng, Brian Caffo, Jeff Leek
I have somehow mixed feelings towards MOOCs provided by Coursera. Their offer is wider than other providers, generally speaking the quality of the teaching is good, and to be honest it is difficult to complain with somebody who is teaching you something for free. On the other hand, the support you receive from the instructors is generally lower as compared to edX, and the same goes for technical support. Also, as I will explain below, their certificate track seems to me quite problematic, as basically trainees have endless attempt to complete the mid-term exercises, and the grading of the final exam is based on peer reviews. This lead to some inconveniences that eventually persuaded me not to complete the specialization.
The full specialization consists of nine courses and a final capstone project. The best part of the corse is probably the first half, focusing on R and github. I do believe that in fact this is a very good introduction to R and a good starting point for the language, especially the three courses Getting and cleaning Data, Exploratory Data Analysis and Reproducible Research. The very first course, The Data Scientist’s toolbox, is a very basic introduction to the topic and it shouldn’t take more than a couple of days to complete for anyone with some computer literacy, and it is hard to consider it a proper course. When it comes to statistics, with the Statistical Inference and Regression Models, the topics are quite compressed in a short time, to the effect that the explanations are carried without great depth. I think I would have struggled much more to complete them, had this been my first introduction to the topic and had I not previously completed a stats course on edX . This said, if you already have a basic understanding of what’s at stake, or a good basis in linear algebra you shouldn’t face big issues.
However, and here comes the trouble, don’t expect much in terms of support, other than the materials provided. This proved to be a huge problem when it came to graduation of the final exam, as you are supposed to be graded by peers that have more or less your same familiarity with a topic. Consider that three or four peers reviews are enough to get graded, and you can easily see how this can create issues, like misleading suggestions if not incorrect grades, especially when it comes to relatively complex problems.
Final line: because of this (and because I was deployed to South Sudan with a terrible connection and tight deadlines), I didn’t feel like spending more money and completing the specialization. And if you are interested in Data science, perhaps this can be a good starting point for R and of course you can take the courses as an auditor, but if you are looking for support and practice, you might want to look elsewhere.