Teaching and Learning Data-Driven Machine Learning With Educationally Designed Jupyter Notebooks

Abstract

This study examines modelling with machine learning. In the context of a yearlong data science course, the study explores how upper secondary students apply machine learning with Jupyter Notebooks and document the modelling process as a computational essay incorporating the different steps of the CRISP-DM cycle. The students’ work is based on a teaching module about decision trees in machine learning and a worked example of such a modelling process. The study outlines the students’ performance in carrying out the machine learning technically and reasoning about bias in the data, different data preparation steps, the application context, and the resulting decision model. Furthermore, the context of the study and the theoretical backgrounds are presented.

Jupyter Notebooks for Teaching, Learning, and Doing Data Science

“…To answer the research question, we investigate three computational essays created by students, which were already analyzed (Fleischer et al, 2022). Further evidence is drawn from recordings of the students' oral presentations of their essays and subsequent interviews regarding their experiences using worked examples and creating computational essays.…”

“…A more detailed description of the data science course and the worked example is given in (Fleischer et al, 2022). The authors analyzed the same data regarding the second computational essay and found that the worked example provided good support for technically creating a competitive decision tree using machine learning.…”

“…However, the reasons for these narrative issues remained unclear. Fleischer et al (2022) hypothesized there could either be a lack of understanding or of adequate documentation skills. Further insights into which issues occurred and why this happened may reveal recommendations for the design of the worked examples (or the teaching module), especially regarding the form of support.…”

“…In the Project Data science and Big Data in School (ProDaBi, www.prodabi.de/en) project, we are developing teaching modules, together with respective tools that students can use to get engaged with data and machine learning models. Findings by Fleischer et al (2022) as well as by Hüsing and Podworny (2022) indicate that creating computational essays is a promising method to let students do data explorations and create data-based decision models while documenting them in the same document.…”

“…Building on insights regarding the use of worked examples in order to create computational essays (Fleischer et al, 2022), in this paper, we want to analyze difficulties in creating computational essays guided by worked examples. We want to investigate how worked examples can be designed to support the creation of computational essays for data exploration and machine learning processes.…”