Duke Data Analysis and Statistical Inference Review

Duke’s Data Analysis and Statistical Inference on Coursera is an introduction to statistics with an optional computational component using the R programming language. The course runs about 8 weeks and covers a considerable amount of ground in that time. It starts with the basics of data and data collection methods but quickly moves on to cover probability, the normal distribution, the binomial distribution, hypothesis testing, confidence intervals, Z and T statistics, ANOVA and Chi squared tests and linear regression. The course is a bit of a whirlwind tour that packs a lot into each lecture. The PDF slides that go along with the videos are a great resource to review the information dumped in each lecture. Many students complained that the course requires more time than the original estimated amount of around 6-8 hours per week. The course was later updated with an estimate of 8-10 hours per week, which is on the conservative side. If you come in with some prior knowledge of stats and R you can get through in 3-5 hours per week.


The professor is engaging and does a good job going through the material while providing adequate face time. The slides are very informative and the video quality is excellent. There are periodic in-lecture quizzes that help test your understanding of the material as you go along. I felt that the frequency of in-lecture quizzes was just about right in this course.


Grading is based on performance on weekly quizzes one midterm and one final exam. You need a cumulative grade of 80 percent or more to get a certificate and you only have 1 attempt on the exams, so it is a bit harder to earn a certificate in this course than it is for most MOOCs. If you choose to go the computational route, a portion of your grade is based on 8 programming labs using the R programming language. You can do the labs on your own or use a convenient web-based programming environment provided by the instructor. The labs provide a basic introduction to R and each one explores some of the concepts introduced in the lectures. The labs take about 30 minutes to an hour and a half depending on your level of experience with programming and R.


In the computational track you’ll also complete a final project involving a statistical analysis of two variables, either from a data set provided by the instructor or a data set you find on your own. The project lets you use the concepts you’ve learned both in class and in lecture on your own. I suspect the project is a bit intimidating to those who are new to R because it involves more computation than the labs and you don’t have the training wheels that the labs provide. The project grade is based on the median score of 3 or more peer assessments.


This is a great course for anyone looking to learn statistics that moves fast enough not to bore those who know a bit of statistics coming into the course.


I give this course is 5/5 stars: Excellent.