Good

good

Wow!

good

ok

.

There is no doubt that Brian is extremely sharp and knowledgeable about statistical inference subjects. However, I tended to agree the following forum comments from another fellow student.

“…This is unfortunately the worst lecture that I have come across in the Data Science stream so far. The presenter zips through it at a lightening pace. The pace, content, presentation, examples - NOTHING - is suitable for the intended audience (i.e. people taking up the data science stream). The lecture appears to have been recorded for some other audience - maybe people taking up a university course in advanced sats - and just plugged-in into this coursera stream. I wish the course publishers (Johns Hopkins) had put in a little bit of thought and effort into this module and tailored it for this specialization stream rather than lazily lifting and dropping a pre-recorded lecture from elsewhere. It should have been slower-paced - maybe split into 2 or more slower-paced lectures that are gentler on the Data Science stream new bees...”

This course gave me a pretty good basis for a lot of subtopics in the field.

However, it was very frustrating at times. Brian, the instructor, clearly operates at a high level and in his explanations he makes frequent conceptual leaps that are difficult to follow. I spent a lot of time on Wikipedia and Googling the details of the various distributions.

Also, the exercises in the associated mini-textbook have a similar problem: there is not enough practice of the basic concepts to give a beginner confidence (no pun intended). Many of the exercises suddenly introduce assumptions that a beginner would never have made. So, expect to get everything wrong the first time, and then look at the answer and figure out the two or three new aspects of statistics that you needed to know.

I have done PhD level statistics courses before, but found that they either went too deep into theoretical mathematics as to completely loose the audience (at least me), or to not even try explaining what is going on under the hood of R or SPSS. What I really like about this cource is it pushed me to do calculations by hand, which really helped me understand the concepts. Dr Caffo is clearly a skilled statistician and the course is at its best when he goes off script (at least off slides) to explain and illustrate concepts. Minus one star because unfortunately the presentation of the material is uneven and some times distracting, e.g. talking very fast.

This was the toughest of all the Data Science courses so far. Without the statistical inference book, the practical exercises and the swirls it would have been very challenging to pass the course. These were very valuable tools. The videos that are available when I couldn't get a practical exercise right, also helped me a lot.

I majored in mathematical statistics 40 years ago, but have never used it in my whole career. But in my opinion this course explained the concepts much better than how it was done 40 years ago. Congratulations to everybody involved.

I have learned so much and was really proud of myself when I passed this one.

I think overall Professor Caffo does a fairly good job of explaining the material, though he covers a lot of topics quickly within the course. So I found myself having to pause and rewind often when taking notes. The course project was definitely the most challenging aspect of the course for me, since my background is in SAS and am an R novice. I spend a full weekend putting the report together. Since so many topics were covered in such a quick fashion, I don't think I retained it as well as I should. Luckily my undergrad statistics background helped a bit, even though it was over 20 years ago.

I get that the subject matter is hard and so this isn't going to be easy to absorb regardless of how it is taught. My biggest challenge was too many overloaded sentences where to understand the new area that was being focused on there was an assumption that i already had firm grasp on the set of other subject areas referenced in the same sentence. A lot of confusion as well arising from terms that sound the same except for one word or context of their use which maybe could be helped along by a summary slide of terms and meaning at the end of each lecture or section.

Nice course with an appropriate level covered for the data science specialisation (assuming people taking these courses very have different prior knowledge of statistics). It would however be good to add a second statistics course to the stream with some more advanced topics. Yet, it is still one of the harder courses of the specialisation.

The only big criticism I have is that the course feels a lot less polished than other parts of the specialisation. It feels like cut and pasted parts of other courses added into one course than its own entity.

I am interested in taking the Regression Models course and took this one as a refreshment for the statistics knowledge I already had. I found the course well done and the resources easy to use and throughout.

As a negative point I would mention that as the topics get more involved, the time dedicated to each one seems to decrease, to the point where both MULTIPLE TESTING and GROUP COMPARISONS are pretty much a briefing, which kind of defeates the purpose of including them on the course in the first place...

I give the course a 8.5 out of 10.

Very... practical approach. Almost no math, almost no theory - but much R code to write and understand. Now almost any person could be a statistician.

I would prefer more details about caveats of the job, at least about p-value controversy. I would prefer more theory. But perhaps it just mean that I should take another course in addition to that one.

Thanks to Brian, Roger and Jeff for teaching me. Thanks to my fellow students - it was nice to interact with you. Thanks to Coursera and to Johns Hopkins University for making this happen.

I know this is a very shallow dive into the field, yet I have the feeling I could have learned more. The absence of judgment whether my calculations in the project were correct or not was confusing too me since in statistics it's more important to correctly interprete the result then to get the result. Calculation and interpretation of results should become routine and more different, practical, coding exercises would be very cool to have. Having that said, I still much appreciate the Data Science Specialization as a whole.

Since I already had some background on probability and statistics, there was a lot I already knew here. However, the use of R throughout and the contents of week 4 (power, multiple comparison, bootstrapping) were all new and I really enjoyed learning them.

The lecturer was really knowledgeable, but I believe sometimes he was a bit monotonic making it slightly harder to follow. However, the lecturer did make a huge effort in explaining everything carefully.

Overall, I really enjoyed this course and I would recommend it.

I liked this course, and found the lectures interesting and would give them perfect marks on being instructive. However, I found that there are some speling errors here and there, and a remarkable frequency of sirens audible in the lectures (is JHU in a difficult neighborhood?), and sometimes it was hard to recover the plots from the git repository.

But, more importantly the curriculum and teach was very good, and the use of examples after the theory makes it much easier to grasp difficult concepts.

Very good and informative course. Brian's teaching style is not the best. The quizzes were not easy. But then this is a University Level course and students need to be resourceful. The textbook helps a lot as does the rmd. files available through GitHub. The mentors in the forum are very helpful.

My recommendation : add one more course project (or two). These really help with learning how to deal with real data and how to apply what we've learned in the lectures.

The course on statistical inference is a crash course on frequentist statistics; in my opinion the contents are appropriate for statistical inference, though most of the concepts that are taught deserve more attention, or perhaps split throughout other courses.

Nevertheless, Brian makes a terrific job to compress all of these topics into a single month course which is impressive - for someone familiar with statistics it's actually an enjoyable course.

I thought the course material was great but I think the grading criteria for the assignment should be more rigorous and check for proper methodology/application of techniques, just because I'd like to know that my approach to a given data analysis is sound and that the conclusions I draw from running tests, p-value adjustment, calculating power, etc. are statistically valid.

Overall a great primer! Thanks to Brian & Jeff for putting this together :)

If you are a beginner and you are just trying to learn statistics then this shouldn't be your first course. It is an intermediate course and if you're just trying to learn stats for fun or basics of it, just to understand the data you are looking at, this course is not it and if you know a little stats and are familiar with some of the concepts and understand the lingo then this course has a lot to offer and you'll learn a lot of fun things.

Very challenging and very valuable as a learning experience. I really liked it and fortunately I passed. I did think it was a bit too fast-paced in some places though. It is so much theoretical and mathematical stuff that it would have been all right to spend 2 months rather than 1, and learn a little bit more in-depth with a little more space for practice and assignments.

Overall a good course. Swift library was used well in order to review ideas presented in the lectures. I feel as though not enough emphasis is put on the importance of multiple comparisons and too much effort is made to shy away from the mathematics, of which there is none. I would give 5 stars if the course was not so mathematically barren.

Relatively, this is one of the best courses and lecturers of the specialization, Brian delivers clear, thorough and well-paced lectures. These lectures on statistics, regression and machine learning are where the rubber hits the road after a lot of prep work to learn R and principles/tools of data science taught in earlier classes.