Mongolia Embraces an Intense Week of Data Science
I’m always on the lookout for new places to take my instructional odyssey. So when I had a student from Mongolia in my Stanford class last year, she introduced me to Badruun Gardi, a Stanford alum and well-connected figure in the main city of Ulaanbaatar. Badruun was confident that Mongolia would be a suitable destination, and he was right!
It wasn’t my first time in Mongolia — my family still considers our trip to western Mongolia 12 years ago to be one of our more unusual adventures. So although I would be spending most of my time in Ulaanbaatar, I was still excited to go back. A few weeks beforehand I learned that a recent president of Mongolia, Tsakhiagiin Elbegdorj, was spending some time at Stanford. I invited him for lunch and our conversation bolstered my anticipation.
My teaching was hosted officially by GerHub, a nonprofit co-founded by my host some years ago. The organization has little connection to data science, but Badruun was enthusiastic about the benefits of my visit and mobilized the GerHub staff and summer interns to organize it — they did a terrific job. Any lack of technical background in the GerHub hosts was made up for by the three Mongolian course assistants they recruited: a current Stanford student along with recent Harvard and Princeton graduates, all sufficiently well-versed in the topics I covered to provide excellent course support.
Design thinking already has a foothold in Mongolia, so we decided to devote my entire visit to data science. Teaching data science over five full days is a luxury, allowing me to cover pretty much all of the material that I have on offer and not be in a rush. (I so appreciated the not-rushing that I’ll likely slow down in future locations, even if it means cutting back on material covered.) The students requested an “ask me anything” session, so I ended the formal class early one day and answered a broad range of questions they submitted in advance; it was a fun diversion for all of us.
There were some unusual and endearing attributes of the 150 participants, who were mostly college students along with a handful of high schoolers and working professionals. Most notable, about 2/3 of them were women. Apparently that wasn’t a surprise: rural nomad families, which constitute about 30% of the Mongolian population, prefer to keep their boys at home for herding while sending their girls off to college. There was very little attrition over the week. Students had excellent attention spans throughout the eight-hour days and were dedicated to learning as much as possible. They even preferred to keep the breaks and lunches short in order to cover more material, which is quite unusual (and hard work for me!) — only in the Philippines have I experienced similar diligence.
Somewhat less unusual was their eagerness to win the Stanford swag given to the first few students finishing each set of assigned problems. Some students were working ahead then sprinting to the front to show me their solutions; one time a nasty accident that may have cost a laptop or two was barely avoided. I have a “one prize per person per day” policy, and a few students realized if they won prizes early each day they could aim to collect all five items on offer: two different colors of string bag, two different colors of sunglasses, and a pen. I never did find out how many grand slams were achieved.
For data science teaching and exercises I use three example datasets: European cities weather, the 2010 soccer world cup, and passengers on the Titanic. Of the 24 countries I’ve been to so far, Mongolia was the first where many of students didn’t know how the soccer world cup works, or much about soccer at all. There’s a national team but it’s not widely followed and unlikely to make a world cup. Perhaps more relevant, kids in Mongolia don’t play soccer growing up — there aren’t many soccer fields and it’s just not in the culture. (Basketball is more popular, and horse-riding.) Once I explained how the tournament’s group and knockout rounds work, the four different player positions, and what yellow and red cards mean, we were pretty much good to go.
No blog or other discussion these days would be complete without some mention of ChatGPT. In addition to briefly covering the topic of generative AI from a technical perspective, I explicitly allowed students to use it for assigned problems if they wished — I was curious what would happen. Only a few tried it, but it did pretty well, particularly for exercises involving the database language SQL. Some know-how was still required to finalize the answers, and ChatGPT made some of the same subtle mistakes students tend to make. I didn’t need this experience to convince me that generative AI is a seismic shift for computer programming and for education, but I was certainly intrigued to see a taste of it first-hand.
It would have been a shame for me to travel to Mongolia and only experience the capital city. After teaching and before a Korea stopover for Stanford business, I embarked on a three-day private tour arranged through Goyo Travel. My terrific guide Uugan-Erdene (“Eric”) and I headed west from Ulaanbaatar in a sturdy Toyota Landcruiser for an enjoyable potpourri of hiking, mountain biking, a local Naadam festival, overnights in yurts including one with a nomadic family, and plenty of driving as demanded by the vast Mongolian steppe.
Next: In September I’ll teach data science short-courses at the African Leadership Academy in South Africa and the related African Leadership University in Rwanda.
