j

This is a great course and I have learned a lot. The teacher is extremely knowledgeable and formulates things very clearly. However, this is really a math course. For me it was hard to stay motivated because the language of the course is mathematics, the teacher juggles with the concepts that my mind was still trying to process and absorb. I was able to finish all exercises, including the honors ones, but when I finished the week 3, I had to redo it completely again and buy a book on Bayesian statistics by John Kruschke which helped me immensely to rethink the basic concepts again. This course could be excellent if it included more reiterations of concepts, was explained in more general language, the pace was slower and most importantly included more practical applications. The typical statistical examples of coin flipping are fun, but too abstract. In the end, I want to know how I can apply Bayesian statistics. A lot of knowledge of mathematics was assumed and I had to look up a lot of concepts myself. The derivations sometimes also went too quick and supplementary materials were quite dense. I think this course is a perfect refresher course for someone who has mathematical background and has taken a Bayesian statistics course some time ago. But for the beginner with some mathematical background (I am familiar with the frequentist statistics, machine learning, calculus) it was too much of a challenge. If it were not a Coursera course, where I can rewind endlessly and work at my own pace, but a regular university course, there will be p=.9 that I would drop out, while my prior for dropping out would be p=.05

I would have given it 5 stars if some of the materials were covered more in depth (e.g. Jeffrey's prior). It seems like someone can dedicate a lot of time learning about how to apply it in different situations and in some instances I had to hunt for more in depth or simpler explanations for specific subjects (such as Jeffrey's prior) in other sources online. Overall the course is helpful and very useful and very well organized and gives a good amount of extra resources to read on but, I think it can become better if, the instructor did not rush through some of the subjects and spent more time explaining (especially towards the end of the course). The discussion forums help in these types of situations but, there will be a lot of searching dedicated to the specifics you are looking for. Overall an update to the course based on feedback of people that completed the course (from discussion forums) seems necessary. Adding an extra 5-10 minutes to some of the video contents can save the student from hours of research on the internet and confusion (sometimes due to the outside source). I'm not saying one should not spend time learning the material further from outside sources. Just saying the explanation might help avoid the confusion caused by looking into other sources.

The course is great and through, however, it lacks intuitional explanations of many concepts. Thus it is hard to follow sometimes. Also, it requires very decent mathematical background while, in my opinion, most of the viewers are rather economists without strong enough base (luckily I'm with actual mathematics BA). One more point is that I find this course rather unfinished because there is so much more about basic Bayesian statistics to say. For example, one of the most important topics for me and reasons to take the course are BVARs. However, they were not even barely mentioned and the course was cut off with simple regressions (without any clear use of prior/posterior ideas described in the majority of the course). Thus, I think that course is an excellent starting point for those who are really good in Statistics and Theory of Probabilities but do not know anything about Bayesian things. And this course should be definitely followed by some other, more applied one.

There are several things in the course that were able to clear up my understanding. The course instructor responds to more questions than I would have expected as well. The course uses a lot of mathematical notation and it helps to take some time with it but once you get the idea of conjugate priors down you can quickly employ them in your own problems. The course covers conjugate priors for several different likelihoods including the normal distribution and the binomial distribution. Although the derivation of the conjugate priors looks daunting as it is written down, the usage of the priors make Bayesian statistics much easier.

This course uses R and Excel but is not a course in either. Most of the computations that are performed for the quizzes are pretty simple and require little skill in R.

I am glad that I have taken the course and would take another if provided by this instructor. I plan to reference the materials provided in the future whenever I need a refresher.

The course is a great introduction into Bayesian statistic analysis. I particularly liked the detailed explanations of where the parameter formulas came from. Also a great thing, in my opinion, was to write the explanations on the glass instead of just displaying the final results. It kind of provided a sense of interactivity and made the material more digestible for a person with not such a strong background in math. It greatly smoothed the learning curve for me and kept interested and motivated to finish the course. In the end the pace accelerated a bit but was still manageable. Four weeks seems a great duration for such a course - not becoming boring and tiring. Honors tests were quite easy, I'd prefer to have a little more challenge. Overall I'd recommend the course for everyone who wants a quick introduction into Bayesian statistics. It provides a solid background for further studies.

What I liked in the course is that it focuses on examples and solving actual problems. The quantity and the quality of the lectures is great, but what I really missed is written lectures where one can always lookup forgotten things or read details etc. Also, one thing that I think might be added easily is a reference to Mathematica and Maple's routines. I'm using Maple and it took some efforts to get on track. And finally, I think that 4 quizes per week is really too much for working people. It's true that the tests weren't that difficult, but it took me about an hour to do each, so I think 30 mins of lectures vs. 4 hours of quizzes is a bit unfair. Of course, my background in statistics is non-existent so it may be that it took me longer than average. But I think the course material could have been spread over say 6 weeks for lighter load on the students. All best to the team!

The course is good as an introductory course to Bayesian statistics. However, there is no much context and no much explanation for many of the calculations. I think the course would greatly benefit from: before each existing video, make a 2 minute video of a real world situation why that approach is helpful (bus arriving every 10 minutes, etc.) so that all students can connect with the content, then the video as it is now, then a 1 minute video linking the formulas with the real world example. As the course stands now, it is difficult to really follow why the formulas have any application and this only appear in the quizzes (where you are being tested, not learning in a non-stressful situation). I think more context and more examples with the formulas would make this course perfect. As of now, it is way too disconnected from real world, although still worth it.

It was a really well taught class and I enjoyed watching it. Unfortunately I seem to lack some basic understanding, since I am not a statistician. Therefore I had problems following the course and had to do quite a bit of research to do on my own to get long. Still, I managed to get 100% correct on all quizzes and all honours quizzes. So it seems, that if you put in enough effort, you can get 100% on the course without understanding many things. This is not to say, that this course is easy, it took a LOT of effort, but it was possible. I will now investigate further to get all of the basics. Maybe I will come back and take this course as a refresher. Other than that I can whole-heartedly recommend this course. The presented material is very well organized and and presented and Professor Lee is a really good teacher.

A great introduction. I feel like I know a lot more about bayesian statistics now. But I do mostly feel like there is quite a bit I don't know, and while I passed, I feel like there is quite a bit more I need to do to really 'get it'. The professor recommended some books in a discussion forum and I'll be going through some of those next I am sure. I also feel, looking back, I should have had some additional math preparation before starting. The calculus was vaguely familiar but with the pace of the lectures, I felt occasionally lost. I would have found it helpful if there was a quick primer on calculus to know and review at the beginning of the course. All in all great course. Loved the presentation method.

This course provides a solid overview of simple Bayesian models and common distributions used in those models. It also provides an initial understanding of conjugate prior distributions and non-informative prior distributions. The R code used in this course is very simple; easy for a beginner, but perhaps a bit simple if you are already familiar with programming in R and doing commonplace frequentist statistical analyses (regressions, ANOVA, etc). Overall, this course is definitely worth taking if you are interested in Bayesian statistics and need a good place to start. There are quite a bit of videos and supplemental materials which allow for a broadened understanding of the materials. Thanks, Dr. Lee!!

This is a decent course that covers an important topic that I've had a trouble finding good resources for learning about.

Pros: comprehensive coverage of the topic at a high level.

Cons: not enough examples to understand what is talked about in the lectures (especially the continuous data and prior with normal distribution lectures) and to anchor the topic in its practical uses.

I recommend supplementing this course with the MIT OCW 18-05 statistics class (I actually put this on hold and did that then came back).

If this course had a lot more practice problems with fully worked out answers it would help tremendously. I understand a sequel to this class is in the works and I look forward to taking it.

This course provides the student a profound understanding of the statistics behind the bayesian approach. Also, it gives some intuition for the difference between the frequentist and the bayesian approach, although that part could have been more explicit in my opinion. It would be very helpful to have more examples on the differences in using freq vs bayesian approach, the gains from using bayesian approach, examples of where the freq approach is limiting / misleading in its 'objectiveness'. More 'real life' examples instead of coin flipping examples - although easy to follow - would be very helpful as well, maybe in a consecutive course with applied bayesian statistics?

An interesting introduction to Bayesian statistics and inference. Not for people with no statistical background, as it does assume you are comfortable with various distributions, expectations, variances, etc. and the 'standard' frequentist worldview (including inferential procedures such as linear regression). The material was well explained, and generally well examined, with a mixture of multiple choice understanding questions, and numeric response tasks which also serve as a very basic introduction to R (or Excel if you are crazy). It was good to see the instructor realising that a light shirt was causing problems and switching to a darker one as the videos went on!

I enjoyed the course. My background is mathematics, but not specifically statistics, though I do have a basic understanding of elementary frequentist statistics. My goal was to understand the fundamentals and uses of Bayesian statistics, having attempted that via a couple of textbooks without much success; this time, I got it!

I do have some reservations about the course. Herbie Lee spent a huge amount of time deriving formulas and methods - a few gaps (either hand waving or 'leave it to the student to finish') would have been helpful, I think. This would leave more time for examples and applications. But the course was well worth my time and effort.

This course covers most of the basics in a very good manner. I personally feel, the last week chapters especially regression do not connect the dots between the foundation that was laid and the resources provided were also not very helpful to fill that gap. For e.g I wanted to understand regression from the bayesian context, the session mostly focused on how to do regression in R and the not the internals of how to understand the mechanics behind from the bayesian stand. I will be helpful to introduce some content that helps the user to move from univariate normal distribution to multivariate normal distribution and explains some intuition behind them.

The course itself is wonderful, and the contents are very thoughtfully selected. I'm not a particular fan of the mirror-technique they use to shoot the videos. Basically, Professor Lee stands in front of a mirror and writes onto the mirror with text markers. On the video you see both him, and the text he writes.

His body often covers the text and generally, it is hard to read. Personally, I see no need to see the professor. Rather, I would prefer a note-taking app (white background). There, old formulas could also be replaced by LaTeX text making everything much more readable, plus there would be downloadable lecture slides automatically.

I liked that the teacher put things into perspective and showed the connections between the different concepts. I deduct 1 star, because the additional material in rare: Meaning, you have to take notes in the lectures to solve the quizzes and to have something for looking things up. Furthermore, in a few lectures it was difficult to read what the teacher was writing, because he was wearing a shirt with a too bright color. (Sounds funny, but I mean this serious ;-) ) In summary, a great lecture and perfect introduction into the concepts. The quizzes are constructed in a way, that they encourage learning rather than frustration.

The course is nice, the lectures are really clear. Professor Lee is brilliant and he often gives some excellent interpretations of Bayesian results. For example, the classic example of testing for rare diseases is explained in terms of ratio of true positives to all positives. Another example is the explanation of predictive mean for normal models, or the explanation of noninformative priors. They're all clearer than what usually found in many books. The only limit of the course is that it's strictly an introduction, thus very useful topics for applications such as hierarchical models or nonconjugate models are not covered.

It was a very nice course that got more practical towards the end. The only thing I found a little bit confusing is the regression part, without theory videos and with practical outcomes that are exactly the same as frequentist approaches.

Don't be discouraged if you come from a background where integers and derivatives are not usual! I come from psychology and I found it a little bit hard at the beginning, but if you put effort you will get to understand almost everything. As long as you get the idea of where things like formulas are coming from and why are they done that way I think it is enough.

I really liked this course. The material is great and the structure of the course is very well organised. A possible improvement, in my opinion, would be to include more explanatory material or take more time in the videos explaining some concepts or derivations. This is why I have to search for other resources in order to grasp some concepts and I took a lot of time in order to completely grasp all the concepts in this course (roughly 10hours for each week). The last week seems a bit rushed and lacks a bit of explanation in the linear regression, non informative priors and in the normal model.

Good curriculum overall, the course can be difficult for students that don't have a strong background of statistics. I found the video lectures lacking because it was mostly formulas and not much explaining. For intuition I had to consult external sources. Most of the quizzes were well designed and challenged our understanding of the subject. While I don't feel that confident in the subject i did gain a good understanding of the overall idea behing bayesian inference.

My advice would be to provide additional videos that give more insight and intuition behind thess concepts.

Really enjoyed weeks 1-3 of the course. It was well done and I felt like I had a good grasp of the materials, and the tests reflected that. The lecturer gave good intuition of what was going on with the math. Week 4 on the other hand was a bit hectic. I didn't feel like I had a good grasp of the material or the underlying math, and lots of it was rushed through. I also didn't feel like the quizzes in week 4 helped me understand the material more. I was able to complete them correctly just by using R, with little understanding of what's going on behind the scenes.

Most of the stuff is explained quite well and I managed to understand it. I am quite satisfied overall and I am glad I completed the course. The exercises, however, were somewhat boring. I wish there were some optional exercises that are more challenging and require you to solve more realistic problems. I also wish there were more additional materials with more in depth theory and examples of how they use these concepts for solving problems that are actually of some use. I feel like these improvements would make the course much more interesting and engaging.

It was my first Bayesian course. Good introduction! However more accent should be placed on intuitive understanding rather than mathematical formalism. To be fair that the issue not only with this course, that the issue with 90% of all stat courses/books. As for me, I find mathematical formalism is hard to digest, intuitive understanding should come first ... May be it's just because of my limited knowledge of stats. I'll update my belief once I get better understanding of stats:) Thank you very much Dr Lee!