The emergency remote teaching caused by the covid-19 pandemic has potentiated the learning gaps of several students in Brazilian education, especially in the K-12 settings. Amidst the many challenges imposed by the pandemic, the adoption of digital tools in the school context has provided the generation of educational data, which can be collected and analyzed in order to provide evidence-based decision making, taking into account all the stakeholders in the teaching and learning process. Such decisions can provide for the personalization of learning, which aims to provide the student with educational resources that promote the building of weakened skills caused by learning gaps. The present thesis plan aims to present the work plan for the development of a Learning Analytics Dashboard tool for teachers in a basic education school in order to support data-driven pedagogical decision-making and to enable personalized monitoring of learning pacting, mainly, the teaching and learning process [1]. the virtual environment, adds to the role of the school
1. Introduction
worldwide COVID-19 pandemic, which has been
imMany school institutions, even without the proper time
and resources, had to migrate classes to the digital
world, through educational apps and platforms [
In the scenario of basic education, which comprises
educational indices, but also with concerns related to
the physical and emotional health of its professionals,
students and family members [
Amid so many challenges imposed by the pandemic,
the adoption of digital tools in the school context has
provided the generation of educational data, which can
be collected and analyzed in order to provide
evidencebased decision-making, taking into account all parties.
such dificulties are potentiated, as it comprises one of
the levels of education from kindergarten to high school, interested in the teaching and learning process [8].
Decision making is a task that is part of the daily routine of a
the most important periods for students in this age group, school manager, as well as the teacher who deals directly
ready existed in the context of Brazilian education [6]
teaching scenario, increasing the learning gap that al- tervene, recommend and, above all, improve the quality
the literacy process and the construction of
mathematical skills, that serve as a base throughout their school
career [
Proceedings of the Doctoral Consortium of Seventeenth European [10].Supporting the decision-making of managers, coordinators, teachers and other stakeholders in student learning. There have been applications of LA techniques in the context of basic education, among these applications are those that enable data-based decision making for teachers and other stakeholders, especially with regard to personalization of the teaching and learning process [11]. On the other hand, most eforts in the use of LA are focused on higher education [12] [13] [14], lacking more research and tools that meet the specific demands of basic education schools [15].
Data-based decision-making is an essential action in the school context, taking into account the growing generation of educational data provided by student interactions with digital tools, as well as traditionally existing data, such as grades and attendance [9].Within this context, educational performance indicators also play an impor1.1. Main goal tant role in decision-making with the aim of improving the teaching and learning process. All data and informaDevelop a Learning Analytics tool for teachers of k-12 ed- tion from the various sectors that make up the school ucation school in order to support pedagogical decision- can and should be used in order to provide insights and making based on data and enable personalized monitor- support decision-making in a timely manner with the ing of learning. aim of promoting personalization and learning recovery.
• Collect studies that used Learning Analytics in 2.1.1. Personalized monitoring of learning in the context of basic education; blended learning • Identify the demands of an elementary school to Due to the global pandemic of covid-19, schools had to adopt Learning Analytics at an institutional level; adapt the way they interact with students, starting to use • Investigate the main problems and challenges as- more disruptive teaching-learning tools and strategies sociated with the adoption of Learning Analytics to guarantee remote classes, such as Google Meet and in the context of elementary schools; Zoom, which were used. to enable synchronous classes • Design of a Learning Analytics tool to monitor [16]. With regard to educational strategies, there was a students’ learning progress; massive adoption of hybrid teaching models, since, soon • Conduct an evaluation of the adoption of the pro- after the beginning of the vaccination period for health posed tool in a primary school and verify if it professionals, many schools began to partially return to supports teachers in how to deal with student the face-to-face model, adopting this approach [17]. learning gaps through personalization of teach- Blended learning is “a model of formal education that ing. is characterized by merging two modes of teaching: traditional and online, also valuing interaction and collective and collaborative learning” [18]. In the systematic review conducted by [19], where they investigated what types of hybrid learning models exist, having found six types. The following models were found: supplementary, inverted classroom, rotational laboratory, study rotation, synchronous collaborative hybrid and dual-collaborative group.
With a diferent approach to the traditional teaching model, blended learning has as its specific characteristic a more personalized learning, respecting the students’ own pace and understanding that people learn in diferent ways [20]. Based on this understanding, it is possible to ofer students learning that addresses their learning gaps and can enable students to learn more individually and efectively. • How can educational data analysis help address
learning gaps? • How to deal with ethical issues in the adoption
of Learning Analytics in k-12 education? • How can learning analytics techniques support the personalized monitoring of student learning in k-12 education? • What is the context of a k-12 education school to adopt Learning Analytics?
that aims to measure, collect, analyze and report data about students and their contexts, as well as understand and optimize learning and the environment in which it ing necessary changes during the school period, in daily takes place [21]. practice [28].
Despite substantial growth in the application of learn- In view of the objective of formative assessment, which ing analytics to improve teaching and learning in the aims to accompany students, collecting evidence of their last decade, most of these works have focused on higher learning process, learning analytics techniques can be education issues and contexts [22] [23]. With a wider used to deal with the measurement, collection, analysis adoption of digital educational technologies in primary and reporting of these collected data, enabling teacheducation, recently accelerated by remote emergency ers and other stakeholders valuable information about classes due to Covid-19, there has been a greater aware- students during the construction of their knowledge, takness of using LA to accompany and personalize the learn- ing into account the individual learning pace. Through ing process [24]. these data, it is impossible to make a pedagogical decision
In the systematic review of the literature conducted by based on data, since with the increasing use of educa[11] 42 studies were identified that applied LA techniques tional technologies, in the context of basic education, in the context of basic education, among the approaches more data is generated during student interaction made are, data distillation for human judgment, prediction, ed- possible through formative assessments made available ucational data mining, discovery with models and cluster- by teachers daily. ing. Most of these approaches developed isolated works Support for teachers’ decision-making has gained a in some sector of the school, but did not address the con- lot of notoriety in recent studies in the area of learntext of adopting learning analytics in an institutional way. ing analytics for basic education [29]. And with regard Some aspects must be considered with regard to the use to personalized monitoring of learning, through the inof AL in basic education, an institutional diagnosis must creasing use of digital technologies in basic education, it be taken into account to understand the needs of the is possible to empower teachers to deal with problems context of a particular school, the ethical issues that are arising from lag and learning gaps, and help students to generated must be taken into account. from the use of the recover their learning [30]. data, it is necessary to use more diversified techniques that take into account, mainly, the personalized accompa- 2.2. Related works niment of the learning and, finally, to use explainability techniques (Explainable artificial intelligence) in the algo- In order to verify the importance of the challenges prerithms used to support students, teachers and educational sented in this thesis plan, a systematic review of the litermanagers [11]. ature was carried out, in order to obtain the state of the art of publications that addressed the application of learn2.1.3. Use of Learning Analytics in formative ing analytics in the context of high school, and later it assessment to support data-driven was A survey of studies was carried out, directly from the pedagogical decision making databases, to update these studies, as well as to identify studies that addressed the use of learning analytics in the Assessment has three general functions: diagnose, con- context of basic education as a whole to support teachers trol and classify. These three functions are represented, and/or managers in pedagogical decision-making based respectively, by the types of existing assessments: diag- on in data. Taking this context into account, some studies nostic, formative and summative.The diagnostic evalu- were identified that aimed to address issues similar to ation, according to [25] , “the fundamental objective is this thesis plan. to analyze the situation of each student before starting The work done by [9] uses the various data generated a certain teaching-learning process, to become aware of by educational information systems, such as: learning the starting points, and to adapt the process to the de- management systems or virtual learning environment, tected needs”. The summative assessment, on the other student diary, library system, digital repository, etc. The hand, takes into account all the content taught, usually authors address that due to the use of these digital tools in divided by two months, and at the end of this process, a the school context, there has been a significant increase test is carried out to verify the acquisition of knowledge in the volume and variety of data that can be captured, [26]. stored and analyzed in order to improve student learn
Formative assessment aims to monitor students’ learn- ing and school efectiveness. In this study, they took a ing during classes, in daily activities and is concerned comprehensive approach to the use of learning analytwith ”determining the degree of mastery of a given learn- ics in Bulgarian education, and developed six machine ing task and indicating the part of the task not mastered” learning models to support decision-making based on [27]. Unlike summative assessment, the focus of forma- data from stakeholders in that context. The models were tive assessment is to collect data to reorient the teaching developed to support students, teacher monitors, classand learning process, pointing out its weaknesses, allow- room teachers, administrators, parents and educational inspectors. Four models were evaluated, for students, and teaching practices. monitor teachers, classroom teachers and parents, and Taking into account these approaches, how to use showed promising results. learning analytics techniques in formative assessments,
In the project developed by [31] learning analytics in the context of blended learning, to collect, analyze dashboards were developed to help teachers make quick and report educational data for teachers in a basic educaand efective decisions regarding student learning ac- tion school to monitor the learning process and support tivities in the classroom. The proposed dashboard was the decision-making process. data-driven pedagogical enhanced taking into account the needs of teachers, with decision-making to help students address their learning a user experience and usability suitable for teaching prac- gaps? tice, taking into account the dynamics presented in basic education. An important feature of this study is the provision of information through real-time dashboards 3. Work plan to speed up teachers’ decision making. The dashboard presented for the educational context was originally de- The purpose of this work is to develop a descriptive veloped for the business context, however it was adapted and predictive Learning Analytics Dashboard - LAD, to be used by teachers. The final prototype was evaluated using data collected from Google Classroom and Khan by 9 teachers, and it was found to have a high potential Academy, to support teachers in pedagogical decision to support pedagogical decision-making. As a point of making in the classroom, in order to identify, monitor improvement, the teachers participating in the dashboard and propose interventions to deal with students’ learning evaluation pointed out the need to use data from external gaps. tools, which are already part of the school context.
The work led by [32] addressed an experiment carried 3.1. Submission of the proposal out with five high school teachers, who were monitored The proposal is to use data from computer-based assessduring a school year. Teachers used information pro- ments, with the support of two educational platforms, vided by learning analytics in their classrooms through Google Classroom and Khan Academy, which are used data provided from computer-based assessments. Such to manage classes and activities in the context of online information served as a basis for the planning of classes, and face-to-face classes. which enabled a more individualized and personalized The APIs (Application Programming Interface) proteaching-learning approach. Teachers reported that the vided by both platforms will be used to access and form insights extracted from the data collaborated in their the data repository, which will serve as input for the conteaching practice, highlighting the detailed information struction of the LAD. The availability of descriptive and about each student, task and responses. In the classroom, predictive data analysis through LADs is the most comteachers used such detailed insights to provide feedback mon way to fulfill the Learning Analytics cycle, which to low-performing students and it was found that those has as a crucial objective, in addition to measuring, colstudents who had these learning gaps performed better lecting and analyzing educational data, to provide reports after performing the data intervention. on this data. from student interactions on digital edu
Finally,[33] investigated the role of learning analytics cational platforms, enabling teachers to make evidenceto assess formative assessments, with the aim of using a based pedagogical decision-making [34]. data-driven approach to inform teachers about changes Through the use of LAD, teachers will be able to track in their teaching practices and how they impact the de- student performance in real time on the Khan Academy velopment of student learning. The authors highlighted and Google Classroom platform, as well as have access that one of the most challenging tasks for teachers is de- to predictive results based on student interactions. In the signing, managing and evaluating formative assessments, context of blended learning, using the rotational laboand this is one of the main reasons for not using forma- ratory approach, students participate, in addition to the tive assessments as a form of feedback for students and traditional classroom lesson, they also interact with digifor teachers themselves to adjust their teaching strategies tal devices, where they will have the purpose of continuthroughout the school year. One of the ways to overcome ing the class started in the classroom. Among the most these challenges, according to the authors, was the use of common activities carried out by students are: research learning analytics techniques that were employed in the on the internet, answering online activities, developing study with the purpose of facing such dificulties and pro- individual or collaborative textual productions, etc. viding personalized feedback on a large scale. Briefly, the Students will use Chromebook devices to carry out data collected from formative assessments were analyzed classes through the rotational laboratory approach, using learning analytics and provided recommendations which are notebooks that use the Chrome OS operating that supported students in a self-regulated learning ap- system and are generally used in the school context. With proach, and enabled teachers to reorient their planning the chromebooks, data regarding online activities will be collected through the Khan Academy API, and data regarding student interactions in the classroom will be obtained through the classroom API and Google Chrome’s Sync function, which tracks logs from the browser.
LAD will support the teacher in decision making while students are working on an assignment, in the classroom, and will provide in real time which students will need support and what kind of support will be needed.
In order to support computer-based formative assessment, through Classroom and Khan Academy, for the construction of the LAD, the Learning Analytics - extracted analytics strategy will be used. There are two types of Learning Analytics strategies, the embedded analytics which refers to the data that is used to inform the student and/or adapt tasks to the students’ skill levels without teacher intervention. And extracted analytics refers to the data that is presented for interpretation and provides teachers with information about the learning process and its results, where it is possible to personalize teaching and learning in the classroom [35]. Figure 1 illustrates the diference between the two approaches.
The present work will use applied research as a type of study, which according to [36] are “research aimed at acquiring knowledge with a view to applying it in a specific situation”, where the need to produce knowledge for the application of its results is the motivation to “contribute to practical ends, aiming at the immediate solution of the problem encountered in reality” [37].
For research purposes, it is characterized as descriptive, as it aims to describe the characteristics of certain populations or phenomena and “can also be elaborated with the purpose of identifying relationships between variables” [36]. Its approach will be qualitative for the analysis of research data, according to Gil (2002, p. 133) “qualitative analysis depends on many factors, such as the nature of the data collected, the size of the sample, the research instruments and the theoretical assumptions that guided the investigation” [38].
Regarding the technical procedure, the study is classiifed as action research, which is defined as a type of empirically based research and has a “close association with an action or with the resolution of a collective problem and in which researchers and representative participants of the situation or problem are involved in a cooperative or participatory way” [39].
In order to answer the research questions and achieve the objective of this study, the data collection tools will be the use of forms, observations and interviews with teachers of the Portuguese Language and Mathematics subjects, of elementary school 2, of the Escola Professor Olindina Roriz Dantas, as well as monitoring student performance through summative assessments made available every two months and through formative assessments made available by educational platforms, discussed in the previous topic.
As a methodology for the data mining process, the CRISP-EDM will be adopted, which is a version adapted for the educational context of the consolidated standard of data mining and knowledge discovery aimed at the CRISP-DM industry. CRISP-EDM fully follows the six steps of the original model, but with educational data mining particularities (RAMOS et al., 2020). ção em tempos de pandemia: perspectivas para o [18] H. Staker, M. B. Horn, Classifying k-12 blended ensino da língua materna, fólio-Revista de Letras learning., Innosight institute (2012). 12 (2020). [19] E. P. Schiehl, I. Gasparini, Modelos de ensino [6] K. L. d. Oliveira, E. Boruchovitch, A. A. A. d. San- híbrido: Um mapeamento sistemático da litertos, Reading and school performance in portuguese atura, in: Brazilian Symposium on Computers in and mathematics in elementary school, Paidéia Education (Simpósio Brasileiro de Informática na (Ribeirão Preto) 18 (2008) 531–540. Educação-SBIE), volume 28, 2017, p. 1. [7] M. Buselli, K. P. Estevão, M. F. Sambugari, A im- [20] L. Bacich, A. T. Neto, F. de Mello Trevisani, Ensino portância da alfabetização matemática no ciclo i do híbrido: personalização e tecnologia na educação, ensino fundamental, Revista Eletrônica de Ciências Penso Editora, 2015.
Humanas 3 (2020). [21] G. Siemens, R. S. d. Baker, Learning analytics and [8] K. Schildkamp, W. Kuiper, Data-informed curricu- educational data mining: towards communication lum reform: Which data, what purposes, and pro- and collaboration, in: Proceedings of the 2nd inmoting and hindering factors, Teaching and teacher ternational conference on learning analytics and education 26 (2010) 482–496. knowledge, 2012, pp. 252–254. [9] S. Gaftandzhieva, M. Docheva, R. Doneva, A com- [22] K. C. Li, H. K. Lam, S. S. Lam, A review of learning prehensive approach to learning analytics in bulgar- analytics in educational research, in: International ian school education, Education and Information Conference on Technology in Education, Springer, Technologies 26 (2021) 145–163. 2015, pp. 173–184. [10] G. Siemens, D. Gasevic, Guest editorial-learning [23] M.-R. Sancho, A. Cañabate, F. Sabate, Contextualand knowledge analytics, Journal of Educational izing learning analytics for secondary schools at Technology & Society 15 (2012) 1–2. micro level, in: 2015 international conference on in[11] E. B. de Sousa, B. Alexandre, R. F. Mello, T. P. Fal- teractive collaborative and blended learning (icbl), cão, B. Vesin, D. Gašević, Applications of learning IEEE, 2015, pp. 70–75. analytics in high schools: a systematic literature [24] M. B. Horn, H. Staker, The rise of k-12 blended review, Frontiers in Artificial Intelligence 4 (2021) learning, Innosight institute 5 (2011) 1–17. 737891. [25] B. Oliveras, N. Sanmartí, La lectura como medio [12] P. Leitner, M. Khalil, M. Ebner, Learning analytics para desarrollar el pensamiento crítico, Educación in higher education—a literature review, Learning química 20 (2009) 233–245. analytics: Fundaments, applications, and trends [26] M. Taras, Assessment–summative and formative– (2017) 1–23. some theoretical reflections, British journal of edu[13] H. Waheed, S.-U. Hassan, N. R. Aljohani, M. Wasif, cational studies 53 (2005) 466–478.
A bibliometric perspective of learning analytics re- [27] B. S. Bloom, et al., Handbook on formative and search landscape, Behaviour & Information Tech- summative evaluation of student learning. (1971). nology 37 (2018) 941–957. [28] L. Cortesão, Formas de ensinar, formas de avaliar: [14] A. Charitopoulos, M. Rangoussi, D. Koulouriotis, breve análise de práticas correntes de avaliação, ReOn the use of soft computing methods in educa- organização curricular do ensino básico: avaliação tional data mining and learning analytics research: das aprendizagens: das concepções às novas prátiA review of years 2010–2018, International Jour- cas (2002). nal of Artificial Intelligence in Education 30 (2020) [29] V. Kovanovic, C. Mazziotti, J. Lodge, Learning ana371–430. lytics for primary and secondary schools, Journal [15] C. Cechinel, X. Ochoa, H. Lemos dos Santos, J. B. of Learning Analytics 8 (2021) 1–5.
Carvalho Nunes, V. Rodés, E. Marques Queiroga, [30] B. M. Batubara, The problems of the world of eduMapping learning analytics initiatives in latin amer- cation in the middle of the covid-19 pandemic, Buica, British Journal of Educational Technology 51 dapest International Research and Critics Institute (2020) 892–914. (BIRCI-Journal): Humanities and Social Sciences 4 [16] A. Schleicher, The impact of covid-19 on educa- (2021) 450–457.
tion: Insights from education at a glance 2020, [31] X. Luo, Supporting k-12 teachers’ decision making Retrieved from oecd. org website: https://www. through interactive visualizations: A case study to oecd. org/education/the-impact-of-covid-19-on- improve the usability of a real-time analytic dasheducation-insights-education-at-a-glance-2020. pdf board, 2020.
(2020). [32] W. Admiraal, J. Vermeulen, J. Bulterman-Bos, [17] J. R. R. Lima, A implementação do ensino híbrido no Teaching with learning analytics: how to connect período pós-pandemia, Revista Ibero-Americana de computer-based assessment data with classroom Humanidades, Ciências e Educação 7 (2021) 10–10. instruction?, Technology, Pedagogy and Education 29 (2020) 577–591. [33] R. H. Sagarika, R. Kandakatla, A. Gulhane, Role of learning analytics to evaluate formative assessments: Using a data driven approach to inform changes in teaching practices, Journal of Engineering Education Transformations 34 (2021) 550–556. [34] B. A. Schwendimann, M. J. Rodriguez-Triana,
A. Vozniuk, L. P. Prieto, M. S. Boroujeni, A. Holzer, D. Gillet, P. Dillenbourg, Perceiving learning at a glance: A systematic literature review of learning dashboard research, IEEE Transactions on Learning
Technologies 10 (2016) 30–41. [35] W. Greller, H. Drachsler, Translating learning into numbers: A generic framework for learning analytics, Journal of Educational Technology & Society 15 (2012) 42–57. [36] A. C. Gil, Como elaborar projetos de pesquisa. são paulo: Atlas, 2006. gil, antônio carlos, Como elaborar projetos de pesquisa 5 (2010). [37] B. Básica, B. Complementar, Metodologia científica,
São Paulo: Pearson Prentice Hall (2007). [38] A. C. Gil, et al., Como elaborar projetos de pesquisa,
volume 4, Atlas São Paulo, 2002. [39] M. Thiollent, Metodologia da pesquisa-ação, in:
Metodologia da pesquisa-ação, 1988, pp. 108–108.