While there are references available in the literature regarding learning experiences with Dataset-Based Learning (DBL) approaches, there is a noticeable absence of a standardized model for designing DBL activities. This gap was identified in this work after performing a systematic literature review (SLR). In contrast to other active learning methodologies, the lack of a common framework for the DBL methodology makes it challenging to compare different DBL approaches. This paper highlights the knowledge gap in the methodology for designing DBL activities and aims to provide a common approach for sharing the view and details about what DBL entails in higher education and how to design a DBL activity. Additionally, it illustrates these concepts with three case studies in different engineering fields.
Figure 3: General flow chart to design a DBL activity
This work begins with a SLR that analyses the use of datasets in higher-education active learning activities. For this purpose, documents in the field of education from the databases of Scopus and WOS containing the words “data” and “active learning” were located. From a total of 236 located results, 16 documents were analysed for this SLR. Most of the reviewed documents present case studies that combine the use of datasets with other active methodologies or that take place in laboratory modules. The review articles also highlight the importance of using real-world data, which can be given to the students; obtained using simulations; or gathered by the students themselves using different methods, such as consulting data sources or running experiments. Some of the review papers also state that using datasets in active learning activities is associated with benefits such as improved confidence and enhance learning. Upon examining the literature, we determine that DBL is an active learning approach that can be used as a stand-alone methodology or integrated with other methodologies, incorporating a distinctive element: placing datasets at the core of the activity. The paper continues by establishing a clear definition of a DBL lesson. In this lesson, the dataset provides information and context to the activity, the material to work with, and the solution to the activity that can be extracted from the information contained in the dataset itself. The DBL approach emphasizes the acquisition of concepts through practical engagement with real-world situations. Concepts are presented as a formalization of the actions that are necessary to work with reality. In this sense, the teaching process is quite different from classical ones that usually present concepts as abstract elements that must be set down by concreting them in realistic scenarios. The main characteristics of a dataset, such as its size and complexity, were analysed, since they need to be considered when designing a DBL activity. Moreover, the benefits and challenges of DBL were deduced from the SLR and by combining the advantages of active learning methodologies, studying real-world situations, working with data, and engaging in collaborative learning. Afterwards, a common approach to designing a DBL lesson is presented. The teaching process is divided into three main phases: course scripting, lesson flow, and evaluation and optimization of the learning process. The steps to be followed in each of these phases are described in detail. Next, we present three case studies contextualized in different engineering fields that were designed according to the proposed common approach, and we outline the benefits of using a DBL methodology as opposed to traditional approaches. Then, the paper presents three case studies in which teachers used the proposed common approach for DBL to design their activities. These three case studies help to prove the application of the proposed approach in different fields of engineering and how using a DBL approach presents advantages compared with traditional methodologies. These case studies help to illustrate what a DBL activity is, which we consider essential in this article, because although we found examples of DBL in the SLR, none of them were explicitly identified as DBL. The three cases contribute, in different roles, to the design of DBL activities, each one emphasizing specific aspects of data analysis, interpretation, and application. In the first one, the activity serves as an introduction to data exploration, where students learn to navigate and extract relevant information, developing quantitative analysis skills, and encouraging critical thinking and discussion about the implications of the data in innovation policies, introducing students to the practical applications of data-driven decision making in innovation management. In the second one, the activity exposes students to handling complex datasets with numerous records and fields, preparing them for real-world scenarios where large datasets are common, identifying relevant characteristics, and improving their ability to discern key variables for analysis. Finally, in the third one, the activity provides a hands-on experience in data collection within an experimental setting, emphasizing the importance of accurate and meticulous data gathering; here, students bridge theoretical concepts and real-world data, reinforcing the integration of theoretical understanding and practical application. Altogether, these cases studies cover key aspects, such as data exploration, quantitative analysis, characteristic identification, predictive modelling, and the integration of theory and practice, present in the complexity and demands of data based on the real-world engineering professional scenarios that students will face and for which we are preparing them. The use of real data coming from publicly available datasets or gathered directly by students, such as the ones used in the examples presented, facilitates the connection between the concepts presented and their application. In addition, DBL offers a good framework for multidisciplinary training, because a dataset can provide a context far from the concepts that the activity tries to teach. Also, DBL provides a dynamic approach to teaching concepts that students may consider “boring”. In many cases, employing an active methodology that requires student participation necessitates a careful scripting of the activity to ensure learning outcomes. However, developing a DBL lesson requires time and skills from the teacher responsible for preparing the activity. Developing an activity that follows this approach needs more effort and conceptualization than the use of classical methodologies. In this sense, teachers should evaluate advantages and disadvantages, their availability for the investment of the effort, and of course, their own belief in DBL, because if they do not believe in this approach, the results could be worse than those obtained using a traditional one. The presented approach could be considered a reference model for sharing the view and details about what Dataset-Based Learning is and the experiences of using it. This paper offers a “meeting point” for different active learning proposals to connect Dataset-Based Learning with teachers’ approaches and goals. The objective of the presented approach is to introduce a reference model to design DBL activities, sharing and showing three case studies in different disciplines, describing the details of their design. The study has notable limitations that should be considered in the interpretation of the results. Firstly, the SLR resulted in only 16 documents that were finally reviewed. The number of documents could be increased by including documents that did not explicitly state their methodology as “active learning”. Moreover, three case studies were conducted using the proposed approach; these three cases are presented briefly in this article, but it has to be considered that they were carried out with a relatively small sample size, a total of approximately 60 students. Consequently, the ability to generalize the results to a broader population is restricted. The findings are context-specific and may not be universally applicable. It is imperative to acknowledge that the limited sample size constrains the statistical robustness of the results. Future work should aim to expand the study to include a larger and more diverse sample, facilitating a comprehensive statistical analysis. By doing so, this research could enhance its external validity and offer insights that are more representative of a broader DBL application.
Providing a Common Approach to Designing Dataset-Based Learning Activities Based on a Literature Review. L. DÍAZ-PÉREZ, F.J. LOPEZ-PELLICER, P. BRUFAU, J. LACASTA, R. TRILLO-LADO, J.A. YAGÜE-FABRA, F.J. ZARAZAGA-SORIA. Applied Sciences, 13(23), 12704, 2023.