Teaching Data Science in Developing Countries

Rahim Mahmoudvand, Bu-Ali Sina University.

---

Two years ago, the regional organization for civil registration in Hamedan invited me to present a lecture about vital statistics (birth, death, marriage and divorce). I knew the audience did not consist of statisticians. For about a week, I pored over their yearbooks. Then, instead of showing them detailed statistical analysis which would not have interested them much, I decided to use interesting visualizations in my lecture and produce new comparisons. The audience feedback was great and since then, I have done two more projects for that organization! I also had a similar experience with another regional organization studying social problems in a rural area.

The take-away message: we need to use/teach effective data analytic tools to help us explore knowledge from data. We must prepare ourselves and the next generation of students in new methods in statistics and data science.

In my university, students in a 4-year BSc program in Statistics must take Calculus, Introductory statistics, Probability, Statistical methods, Sampling, Linear algebra, Mathematical statistics, Stochastic processes, Design of experiments, Nonparametric methods, Multivariate analysis, Regression analysis, Time series analysis, Statistical quality control and Statistical computing. Like most developing countries, we have a centralized educational system in which the Ministry of Science identifies the main courses in all programs and the departments are not allowed to change the course curriculum at all. However, we can and do suggest new courses which may be taught after review by a council. We recently proposed exciting new courses such as Data mining, Topics in Applied Statistics, etc. We are also including software such as R in many of our courses. We need courses on Machine Learning, Statistical Consulting, etc. I think it is wonderful that many universities in developing countries are starting to offer similar courses in their Statistics programs.

These days, I am using the problems and data from my projects (vital statistics, social data from rural areas) to show my students how they can use data analysis tools using R, say, when they face real-world problems. I think that many of our students have problems with interpreting data and results and we need to teach a course on consulting and report writing.

By the way, I think simple data analysis tools are wonderful to answer many questions. I remember the famous quote by John Tukey, “An approximate answer to the right problem is worth a good deal more than an exact answer to an approximate problem.”