Data Science is an interdisciplinary eld devoted to extract knowledge from large amounts of data. There is a great variety of programs that address the teaching of this eld with a growing demand of professionals. However, data science pedagogy tends to emphasize general aspects of data and the use of tools instead of the its scienti c dimension. This position paper describes an ongoing educational innovation project for the use of the Challenge-based Learning approach to teach and learn Data Science. In this approach, students work on solving complex and real world problems while the learning is obtained by iterating through three main phases: engage, investigate, and act.
Data science (DS) is an interdisciplinary eld devoted to identify patterns and
extract knowledge by mining large amounts of structured and unstructured data.
Among others, DS includes: machine learning, data processing, statistical
research, and their related methods. This science has become a revolution that
has changed our manner of doing business, health, politics, education and
innovation [
Challenge-based learning (CBL) is a new learning approach created by Apple
Inc. in collaboration with teachers and leaders in the education community. CBL
is \an engaging, multidisciplinary approach that starts with standards-based
content and lets students leverage the technology they use in their daily lives
to solve complex, real-world problems" [
Data science is in a privileged position with respect to other branches of
knowledge to articulate learning through experiences and challenges [
This position paper presents an ongoing educational innovation project
focusing on using the CBL approach in a DS course, as part of a Computational
Biology master degree at the Technical University of Madrid (UPM). Students
will work on challenges at the level of a Kaggle competition with special
preference for active and multidisciplinary problems. Based on the 2016 update for
the CBL framework proposed by Apple Inc. [
The paper outline is as follows: after describing the background of the presented innovation project in section 2, the project details are given in section 3. These include the scope and students' pro le, the project goals, the timeline and educational resources, the evaluation, resulting products, and di usion plan. Section 4 explains the expected contribution to the improvement of learning quality. Finally, section 5 concludes and presents future lines of research and work. 2
The great diversity of applications and the growing demand of experts in the DS
eld has made courses, books and manuals in DS proliferate [
Iris [
We executed an educational innovation project last year, where the
limitations of this standard pedagogical approach were revealed [
EL brings real life experiences into the classroom which must be integrated
with the goals and objectives of the discipline theory [
Some of the results of this previous project [
This section explains the ongoing project details [
The project will be developed with students attending the master in Computational Biology, o ered by the Computer Science School at the UPM. More speci cally, in the module of \Knowledge representation and acquisition".
The pro le of these Computational Biology students is multidisciplinary, belonging part of them to the world of biology and part of them to branches of information technology. Data science and CBL provide an exceptional framework to establish synergies between these two main pro les, e.g. applying computer science techniques to biology or vice-versa. In this vein, one of the requisites of the challenges will be to include both pro les in each work group.
Students will also have the opportunity to apply the knowledge acquired from other master's modules such as \Statistical Analysis and Data Visualization" or \Machine Learning" to real world problems. The results of this project will be directly applicable to other modules, whether of this master degree or of the master in Data Science (EIT Digital Master School)2, and also in courses of the Bachelor Degree in Computer Engineering3 such as \Data Mining". 3.2
The goals of the presented project are the following:
{ G1. Development of methods for CBL in Data Science and Computational
Biology. This goal includes the instantiation of methodologies and general
frameworks of CBL to the speci c eld to be treated. Among these
frameworks, we can point out: the 2016 update for the CBL framework proposed
by Apple Inc [
by the Monterrey Institute of Technology and Higher Education in 2016 [
The phases of the Apple Inc framework are depicted in Fig. 1.
{ G2. Study of speci c challenges in the eld of Data Science and
Computational Biology. Application of the explored, extended, and instantiated
methods in O1 to Data Science. This goal addresses the selection of concrete
challenges and for a speci c student pro le. The students will work on a
challenge at the Kaggle competition level with special preference for active
and multidisciplinary problems. Examples of past competitions that t the
students' pro le are \Predicting a Biological Response", \Merck Molecular
Activity Challenge", \Shelter Animal Outcomes", \Leaf Classi cation", or
\Zoo Animal Classi cation".
{ G3. Integration and documentation of tools for the support of CBL in the
course of "Representation and Acquisition of Knowledge". This goal includes
the analysis and documentation of tools available to support the CBL in this
speci c module. Although Kaggle has a working environment, Kaggle
Kernels4, this framework will be combined with other tools, preferably free and
open source, that meet the needs of the CBL. Among others and according
to the 2016 update for the ABR framework proposed by Apple Inc. [
students will need: a calendar, space for collaboration, and storage of documents. Project management tools such as Trello5 and Asana6 as well as free alternatives will be considered. 3.3
The following three major tasks will be addressed with a clear correspondence to the three goals explained above: { T1. Developing of methods for CBL in Data Science and Computational
Biology. { T2. Analyzing and selecting speci c challenges in the eld of Data Science and Computational Biology. { T3. Integrating and documenting tools for the support of CBL in the Representation and Acquisition of Knowledge.
The project kicks-o on February 15, 2018 and an ends on November 15,
2018, giving 9 months of project numbered from 1 to 9. In an iterative and
incremental approach such as the Scrum methodology [
As shown in Fig. 2, three iterations are considered for each task. This allows the tasks results to feed back to previous tasks. Moreover, there is an overlap into the di erent tasks as expected in an iterative and incremental approach.
The following educational resources will be used: { Scienti c repositories available in the UPM as ScienceDirect7. { The Institutional Teaching Platform of the UPM (Moodle)8. { Data repositories and contests websites in Data Science such as Kaggle. { Resources of the Department of Arti cial Intelligence as web servers.
3.4
The evaluation of the project will be carried out through the generation of an
e-portfolio by the students attending the module under study. The e-portfolio
is a digital collection of evidence which includes: demonstrations, resources, and
achievements obtained by students. According to the CBL framework proposed
by Apple Inc [
{ Final re ections on what was learned. 3.5
This section describes the tangible products resulting from the project (methodological guides, reports, educational resources, etcetera) with a description of their potential for internal and external transfer. The following deliverables will be elaborated: { D1. Report on methods for CBL in Computational Biology and Data Science. { D2. Report on the analysis and selection of appropriate challenges for the learning of Computational Biology and Data Science. { D3. Manual of tools for the support to the CBL in the Representation and
Acquisition of Knowledge. { D4. Report on the evaluation of results based on e-portfolios created by students. { D5. Journal and conference papers for the dissemination of results.
As described in section 3.1 these deliverables have an internal transfer in the UPM to other modules of the master for which it is proposed, other masters, and other degrees. These products can also be a competitive advantage in the organization of massive open online courses (MOOCs) by presenting a pedagogical prescriptions.
The main di usion materials generated in the project will be the deliverables 3, 4 and 5 explained in section 3.5. There will also be contemplated: { the construction of a website that collects all the deliverables, { news for di usion at the UPM, { microblogging posts (Twitter) in the department and the school, { radio interviews to disseminate educational innovation. 4
Thanks to the application of the CBL approach, a signi cant improvement of learning and teaching quality is expected. Among others, this improvement will be re ected in: 1. A deeper understanding of Data Science for Computational Biology, allowing to diagnose and analyze problems before proposing solutions. 2. A greater commitment to involve the student both in the de nition of the Data Science problem to be addressed and in the solution that will be developed to solve it. 3. Development of skills to investigate, create models, materialize them, and work collaboratively and multidisciplinary. 4. A closer approach to the reality of their profession, establishing relationships with specialists in the Kaggle platform that contribute to their professional growth. 5. Strengthening the connection between what they learn in the Master's and what they perceive in the professional world. 6. Development of high-level communication skills, through the use of social tools such as Kaggle forums and media production techniques, to create and share the solutions developed by them.
Moreover, this teaching innovation project [
This position paper has presented an ongoing educational research project to use the challenge-based learning approach for DS in the context of a master in Computational Biology. The paper has revised the background, including a number of shortcomings repeated in current DS courses, and the results obtained in a previous project, based on considering experiential learning for DS. The ongoing project details have been presented: the scope and students' pro les, goals, timeline, teaching resources, evaluation, resulting products, and di usion plan. The project details allow interested professors to extrapolate our case to their speci c environment and audience.
The contribution to the improvement of the teaching quality of the project has also been explored, highlighting a deeper understanding of Data Science for Computational Biology. This allows students to diagnose and analyze problems before proposing solutions. DS is not only about data and tools to manage them as classic DS courses may suggest. DS is more about \science" and the scienti c questions we can answer with data. Therefore, a major advantage in using CBL for DS is that teachers do not present the answers before students ask the scienti c questions by themselves.
Our main future works include to create a survey on the acceptance and quality of challenges proposed, to study new theoretical frameworks for applying CBL in DS, and the exploration (or implementation) of software tools to develop e-portfolios. Moreover, two students have been hired to assist in the search and development of speci c challenges for Computational Biology.
This research work is supported by the Universidad Politecnica de Madrid under the education innovation project \Aprendizaje basado en retos para la Biolog a Computacional y la Ciencia de Datos", code IE1718.1003; and by the the Spanish Ministry of Economy, Indystry and Competitiveness under the R&D project Datos 4.0: Retos y soluciones (TIN2016-78011-C4-4-R, AEI/FEDER, UE).