Project-Based Learning for Scaffolding Data Scientists’ Skills
Fernando Martínez‐Plumed, José Hernández‐Orallo · 2021
Project-based learning (PBL) has become one of the most widely-used instruction methods in Computer Science (CS). Data Science projects can be thought to be similar to CS projects, but there are some key particularities that require a specific account of PBL for data science. PBL has clear benefits for engaging students to apply technical knowledge learned in other courses, address relevant industry problem-solving, provide space for multiple pathways to a given solution, and foster the collaboration with other students developing inter-personal and competitive skills. However, the exploratory character of data science projects, which do not start with a clear specification of what to do, but some data to analyse, present a challenge to the application of PBL. Looking for value in the data, building novel narratives and finding new insights is quite often perceived as unsystematic from the point of view of the students. Students get anxious soon. However, giving scaffolding as a step-by-step guide of what to do next is completely against the skills about curiosity, innovation and autonomy that a data scientist should possess. In this paper we introduce the characteristics of a standardised rubric that emphasises the value, the innovation and the narratives, and places it as the main scaffolding structure for the course. From the results of a PBL data science course at the MSc level, we see how the exploratory character of data science, as well as the range of proactive, curious and inquisitive skills that data scientist should have validates the application of this PBL design in data science courses.