Measuring students’ progress in Machine Learning

Abstract

Machine Learning (ML) is a rapidly growing field, therefore ensuring that students deeply understand such concepts is of key importance in order to certify that they are prepared for the challenges and opportunities of the future workforce. Despite this, literature on teaching ML and assessing students' understanding with regard to this field is scarce. Hence, this research aims to provide an extensive analysis of the best practice within the ML field, with the main focus of the study being the decision trees and random forests classifiers. An analysis of learning outcomes is conducted using Bloom's taxonomy, guidelines for creating assessments that reflect students' understanding levels are provided and a series of interviews and surveys are conducted in order to analyze the need for certain questions during the course examination. The results are then analyzed and key findings such as the need to structure the course such that decision trees are assessed as a prerequisite for learning random forests are further discussed. The research is concluded with a set of recommendations that could be integrated into future editions of the course in order to assess student progress in a more efficient manner.