Paris hosted its first-ever master class, math+econ+code Masterclass on Optimal Transport, Choice and Matching Models, delivered by Alfred Galichon, NYU Paris site director, and professor, Department of Economics and Mathematics, Courant Institute, in late June.
Interdisciplinary in nature, the five-day course focused on “models of demand, matching models, and optimal transport methods, with various applications pertaining to labor markets, the economics of marriage, industrial organization, matching platforms, networks, and international trade from the crossed perspectives of theory, empirics and computation.”
Interestingly, it is Paris’ world renowned cooking schools, such as Le Cordon Bleu and L’Atelier des Chefs, that inspired Galichon to take a new approach to the delivery of coding lessons. Galichon says his approach was informed by “scientific-context-based learning principles, where theoretical tools are introduced just in time, as needed, at the point in time when they are called for by the specific application.” Under this model, much as culinary students would in a kitchen, Galichon’s students essentially take on the role of apprentices, and over time, move from their position at the periphery of a learning community to the center as they build their expertise.
Galichon explained that the course starts with a presentation of “the raw ingredients to the student, which in this case, means the data, such as the characteristics of consumers and of products.” Following this, he describes to the class “what we will cook, which in this case means the type of matching or pricing problems we will solve.” To make this delivery format possible, Galichon emphasized that “[t]he time-consuming data preparations” – the tasks of the sous chef – “have been done off-line. I show the ideas on the whiteboard, and then I cook/code them myself in front of the students,” explains Galichon. “The students then code themselves, inspired by what they saw.”
Galichon grounds his pedagogy in “a ‘learning-by-coding’ philosophy,” which involves creating code “in front of the students rather than showing them lines of pre-written code.” In this way, students actively learn in-situ, rather than via a scripted method that relies heavily on the passive visualization of code and other course content. Delivering immersive courses, he says, has allowed him to be “quite creative and test new pedagogical ideas that may apply to other courses of this type.”
The curriculum, said Galichon, was constructed to serve the needs of two core groups of doctoral candidates. “There is a big demand from students in economics who look to acquire coding skills, and who want to develop a deeper understanding of the mathematical structure behind the economic models,” he said, “and at the same time from students in math/computer students who seek to understand better the economic applications of their tools.” Crafted around these requirements, the “course provides students with the conceptual tools and coding skills in an apprenticeship philosophy.”
The masterclass quickly attracted “a wonderful mix of students,” said Galichon, from quantitative disciplines, including economics, math, computer science, and engineering. In addition to students from NYU, the inaugural cohort comprised international enrollees from Faculté des sciences de Tunis, and Harvard as well as others from “very strong [local] institutions” such as Sciences Po, and Ecole Polytechnique/Ensae.
The international nature of the masterclass is also part of the linguistic aspects of learning to code. The course presents several different types of coding languages simultaneously, lending to a sense of multiliteracy within the learning environment. This is accomplished, Galichon explained, with “the presence of ‘veteran students’ who, after the lectures, present the course material coded in other programming languages than the one I am presenting.” He added that “These are among the novel ideas that I will test in Paris, which will thus serve as a lab for pedagogical innovation.”