Learner Reviews & Feedback for Inferential Statistical Analysis with Python by University of Michigan

Do you do usability tests of your courses? Like you can test a landing page - you pick a random person to perform a certain action on your landing page, and see where they struggle or what is unclear? If you did this with this course before going live, it would benefit everyone. Right now the quality of this course is too low, concepts are not explained enough, and the assignments (especially week 3) contain wrong instructions and errors.

If you want to learn basic and inferential statistics - I would advise checking out the courses with these name from by University of Amsterdam(you can take them without taking the specialization). they are much clearer. And then if you want examples of Python code - take this course. Just check out the forums first. As of jan2019 the Python Notebook used for the week3 assessment had various problems.

This course attempts to cover very useful topics but falls short on several areas. 1. Multiple errors in the assignments. Practice exercises don't have any answers for students to check. 2. Course slides are not provided. 3. Lack of support to questions asked in forum. I learned a lot from the course but a significant amount of time could have been saved if the issues I mentioned were addressed.

In this course, they cover making confidence intervals and calculating p-values given a specific test scenario (compare sample proportion to population proportion, sample mean to population mean, two sample means to each other, etc). While they go though each statistical procedure clearly, I feel like a lot of underlying context is missing. What is the different between a z- and t-distribution? Why do we use those distributions? How do the different tests relate to each other? Etc. It feels like this course needed an extra 50-60 minutes of lecture time to tie all these concepts together. A textbook to follow along would have been great too.

Alltogether the course was great. I learned so much and understood some principles I did not understand when having read of them before.

However in some notebooks, calculations were wrong or notbooks were missing alltogether (week 4, last jupyter notebook). Furthermore it can be annoying if you cannot trust a result of a statistical analysis in a notebook because there were other mistakes before. That's why I give you "only" 4/5 stars.

Dissapointing. Full of errors during all the course. I came from the first course and I found the second one dissapointing. Some issues I found I thonk they should solve:

*The assessments have no sense. Which is the sense of asking if 0 or 1 corresponds to napping or not napping from a datasheet? In general I also found myself wasting a lot of time trying to understand the definition of columns of NHANES or other datasheets which I think is useless. In overall I found that the assessments do not evaluate the progress of the student.

*The assessments sometimes are not related to the concepts explained during the week (Name that scenario) or in the case of Pythos quiz ask for unexplained functions (week 3).

*Some assignements, like the Chocoloate assignement, are about previously not explained concepts (cross-over test desing).

*The Jupyter Notebook have several errors, reported in the discussion forums but not solved. Moreover, some parts of the code shown are not explained. And finally, the way to solve a problem is different in the explanation video and notebook; this is very confusing for the student.

*Week 3 explains several times the same concept (CI and p_value) but them skip to explain the Power or SampleSize in detail, which are just mentioned in a Jupyter Python notebook. Moreover, I found missing a detailed explained on when to use z_test or t_test; I would had dedicate some time to this explanations, and less to repeat the same CI and p_value examples over and over.

*Week 4, is just a review of week 3. I like the way Brady T. West explains the concepts, but once agains I found extremely repetitive showing so many examples of the CI and p_value.

Great lecture content. Poor quiz design.

Very clear and interesting lectures, but quizzes and Jupyter notebooks could benefit from some additional proofreading and pre-release testing. Material in last week is out of order. Spent a few hours some week just figuring out the mistakes with the help of the course forum.

Also, I would have liked to have a bit more background and explanation, e.g. information on why we using a particular distribution or a particular test, not just how. While a complete derivation of all the material would clearly be out of scope, other courses did a better job of introducing the theory behind their methods.

Good lectures but too little practice and quizzes that don't cover all the material. Very little Python.

No lecture slides or "handouts" to summarize procedures or formulae that tend to jumble together for the various scenarios you learn. Some of the lectures told us to find tables needed to do the quizzes online, no more specifications. That was very disappointing.

This course is a good statistics course, but a poor Python course. Python is practically an after thought in each week's lesson as the focus in the lecturing learning methods is entirely verbal rather than supported by in lecture use of Python. The Python review at the end of each week before the assessment is not connected enough with the lecture materials and makes for a very disjointed week of learning.

The course contents are good to an introduction or refreshing in statistics but the assigments are not really well prepared, and contains many unrepaired errors. This drops down the level an educational potential of this course (and the entire specialization) and converts it in a poor educational resource and a waste of time, in my opinion

I must say that this is a must take course for ones who are aspiring a career in Data Science. All the concepts were laid out so beautifully and it was explained very clearly with visualisations of each real-life-examples. I enrolled in this specialisation before starting my Machine Learning so that I have all the necessary fundamentals of Statistics. Brady Sir & Brendra Ma'am are simply phenomenal, the way they explain the concepts are incredible. The concepts gets etched in one's memory.

From the introduction, the course is supposed to build the knowledge ground up for a beginner in Statistics. However, it falls short in clearing many concepts and the principals end up being vague in a lot of sense, hence, there is a lack of cohesiveness in the concepts spread across weeks. Fortunately, I took an open course offered by Stanford University of the Inference Concepts explained in this course; before taking up this course.

Overall, this is a good course for someone who is familiar with the Inference concepts. For a beginner, a significant amount effort would be required to catch up to these concepts.

I enjoy the lectures. The professor has a good speaking and teaching style which keeps me interested. Lots of concrete math examples which make it easier to understand. Very good slides which are well formulated and easy to understand

This is a very great course. Statistics by itself is a very powerful tool for solving real world problems. Combine it with the knowledge of Python, there no limit to what you can achieve.

Great Course. There are so many example to understand the topic. I really enjoyed every lesson of this specialization. I am going forward for the next one.

It is absolutely great. Instructors are veeeery pasionated with what they do, and the course material is very good.

I really like this course.

Some concepts not covered in much details - ttest, ztest, one tailed test, two tailed - how to use them,when to use them... one needs to go outside of the course to first understand them. These concepts are not explained but are asked directly in assignments which leads to a lot of confusion.

There are no feedbacks for wrong answers in QUIZ, of course the course and the quiz are not at all relatable, lol

This "Inferential Statistical Analysis with Python" course went in depth into the topics of confidence interval and hypothesis testing in ways that were not covered in school. You will learn how to perform hypothesis tests for key areas such as : 1. (Single) population mean 2. (Single) population proportion 3. Difference in proportions 4. Difference in means for paired data 5. Difference in means for independent observations/groups The course covered lecture videos, well-prepared readings, Jupyter notebooks to introduce concepts as well as practice notebooks, lab walkthroughs, written assignment and quizzes. Brady may speak alittle too fast, especially when it comes to long sentences, so you may need to rewind certain segments of the videos numerous times to revisit some concepts as you reflect and learn. Through the Jupyter notebooks, you will better understand challenges when working with datasets and the nuances of how to perform these hypothesis tests in real life, compared to "sampling design and hypothesis testing" on paper (back in school in the past). The discussions forums was also actively monitored by a TA who got back to me usually within a day, which helped to unblock conceptual roadblocks quickly. You may encounter some issues with the Coursera platform: 1. Labs may fail to load at times, even after following instructions to restart the machine which your lab runs from 2. Your discussion forum comments may disappear right after you post. Remember to copy your comment somewhere else (e.g. on a notepad) as you may need to refresh the page and post your comment again.

They could increase the rigor mathematically on this OR spend more time on creative code. Code was great. Math was very very easy so you don't have to listen to the videos if you don't want and just grab the lecture slides. I find on many of these courses they are either way too rigorous or not rigorous at all where the lectures are a bit of waste and you can just read the books/slides (Sounds like Undergrad math/stats lectures all over again? aha)

HIGHLY Recommend course if anything for the CODE in the notebooks. If you're just getting back up to speed on Pandas I found that these helped with sorting the data and reminding you of the difference of built in methods for stats with pandas and numpy.

Laslty, I did not like that only 3 people grading the assignment if we are going to do this crowd sourced grading. I found one grader in particular didn't understand what a p-value was and marked me down and another from the previous course didn't have a handle on english enough to understand what I was writing on my memo.

That said thank you for the course.

I found it a lot more interesting than the first one. Better examples, better structure.

My only complain: I do realize that week 4 was meant to be as an overview of the course, but... the structure of week 4 is much much more clear than for the 1-3 week. I for one, wish i watched week 4 first.

Why i say that ... because the examples were all over the place... at some point i wasn't sure if i was supposed to use a t-test or a z-test. Some examples were repeating themselves throughout the course. Sometimes the code presented in the video was not the same as the one present in the lab files.

I can say that at the end of first 3 weeks...i understood almost everything there is about the inferential aspect of statistics...but... the python part was a complete mess. After week 4 & this: Statistics 350 – Formula Sheet (umich.edu) ... everything became clear. The link should be provided in the course... I stumbled on it by accident in the forum section.

This course went into great detail about how to calculate confidence intervals and p-values, what these mean statistically, and some cautionary tales and misconceptions regarding them. It was really repetitive, which I appreciated, as it really drills the concepts in, but if you really catch on quickly to mathematical concepts you might find this a bit annoying. The applications with Python are quite useful - the course goes into detail not only with how to manually input values, but encourages you to write your own functions and ultimately access the statsmodels library to simplify this. It's nice to know what's going on under the hood in the libraries we so often depend on.

Fantastic! I really liked how the course showed the groups of tests (one test, paired test, difference) and then bucketed it by proportions vs means. I also really liked how the course tied hypothesis testing with confidence intervals to really get a sense of how it all ties together.

The python piece is good. I have some experience with python and pandas, so I was ok. Not sure how easy it would be if you have never worked with python before. Then again, the creators of the course state that you should have some familiarity with python.

I really enjoyed learning through this course. The instructors manage to explain in great clarity the concepts that it covers, and they are also very engaging and fun to watch. Professor West is also making a great effort paying attention to details that every sound statistical analysis should follow but is often overlooked. The Python notebooks are very informative too, offering the right amount of challenge to learners. Highly recommended course overall.