ARLA Model and
4 FUNDAMENTALS
4.4 Learning Analytics
4.4.2 Differentiation
As discussed above, LA is closely related to EDM, information visualization, and academic analytics (sections 4.3.2 to 4.3.4). Duval and Verbert (2012) also mention a connection between LA, big data, data mining, and information visualization:
“In this domain, huge data repositories collect traces of where people go, whom they interact with, what they buy, etc. Analytical applications then try to make sense of the data, either algorithmically through data mining techniques, or through information visualization techniques in visual analytics.” (Duval and Verbert 2012)
Table 5. Definitions of LA.
Reference LA Definition
1st Conference on LAK (2011) and also adopted by (SoLAR 2013b) and (Wikipedia 2013a)
„Learning analytics is the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs“
(Elias 2011) „Learning analytics is an emerging field in which
sophisticated analytic tools are used to improve learning and education. It draws from, and is closely tied to, a series of other fields of study including business intelligence, web analytics, academic analytics, educational data mining, and action analytics. “ (Siemens 2010) „Learning analytics is the use of intelligent data,
learner-produced data, and analysis models to discover information and social connections, and to predict and advise on learning"
EDUCAUSE Next Generation learning initiative; as cited in (Siemens 2010)
Learning analytics is “the use of data and models to predict student progress and performance, and the ability to act on that information”
Horizon Report 2011 (L. Johnson et al. 2011)
„Learning analytics refers to the interpretation of a wide range of data produced by and gathered on behalf of students in order to assess academic progress, predict future performance, and spot potential issues. Data are collected from explicit student actions, such as
completing assignments and taking exams, and from tacit actions, including online social interactions, extracurricular activities, posts on discussion forums, and other activities that are not directly assessed as part of the student’s educational progress. Analysis models that process and display the data assist faculty members and school personnel in interpretation. The goal of learning analytics is to enable teachers and schools to tailor educational opportunities to each student’s level of need and ability." (p. 28)
Ferguson (2013) explains the differences between EDM, LA, and academic analytics by assigning main research questions to each of these fields:
“The emergence of learning analytics as a field in its own right meant that there were now separate groupings focusing on each of the challenges driving analytics research.
• Educational data mining focused on the technical challenge: How can we extract value from these big sets of learning-related data?
• Learning analytics focused on the educational challenge: How can we optimize opportunities for online learning?
• Academic analytics focused on the political/economic challenge: How can we substantially improve learning opportunities and educational results at national or international levels?” (Ferguson (2013), p. 8) Duval and Verbert (2012) see EDM as more focused on automating processes, whereby visualization is more about supporting users awareness and decision processes. They illustrate the distinction between EDM and information visualization with autonomous vehicles that could use algorithms to either “steer the learner in the right direction” or present dashboards that “support people in being better drivers” or rather “help learners to be more aware of what they do, support self-reflection and enable sense making”.
The definition of LA in the Horizon Report of 2012 was quite similar to the definition of the Horizon Report 2011 (see Table 5). But the second part, concerning the objectives had been enhanced to state:
“The goal of learning analytics is to enable teachers and schools to tailor educational opportunities to each student’s level of need and ability in close-to-real time. Learning analytics promises to harness the power of advances in data mining, interpretation, and modeling to improve understandings of teaching and learning, and to tailor education to individual students more effectively. Still in its early stages, learning analytics responds to calls for accountability on campuses and aims to leverage the vast amount of data produced by students in academic activities”. (Johnson, Adams, and Cummins, 2012, p. 26)
This citation highlights timeliness and efficiency of LA. It describes that LA makes use of advances in EDM.
It can be concluded that LA in general builds upon the research findings of several related fields to improve teaching and learning.
4.4.3 Process
Chatti et al. (2012a) presented a three-phase model for the LA process. As illustrated in Figure 3, it is an iterative cycle, which is generally carried out in the
phases: (1) data collection and pre-processing, (2) analytics and action, and (3) post-processing.
Figure 3. LA process. Source, Chatti et al. (2012a).
• Data collection and pre-processing: From a technical perspective, the first step is to collect data. This data comes from various educational environments and systems. Data aggregation and pre-processing is often necessary, since the collected data may be too large or include irrelevant information. Also, it might be helpful to transform the data into another format, which is required for a specific LA method.
• Analytics and action: The next step is to use LA methods for analyzing the data according to the goals of the users. Data can be explored in order to discover hidden patterns. Information visualization techniques are especially useful to help users understand analytics results based on large data sets more quickly. This can support them in their decisions and actions. Taking actions is the primary aim of the whole analytics process. These actions include monitoring, analysis, prediction, intervention, assessment, adaptation, personalization, recommendation, and reflection. • Post-processing: The post-processing phase serves for the continuous
improvement of the analytics exercise. Based on experiences of previous iterations, it may involve the collection of new data from additional data sources, refinement of the data set, identification of new indicators, modification of variables/filters, or selection of a whole new analytics method. Data$collection$ and$pre. processing$ Analytics$and$ action$ Post.processing$
Elias (2011) compares the analytical process with a model of the ‘knowledge continuum’, which – according to her – was used by Baker (2007). It defines four units – data, information, knowledge, and wisdom – which are transferred from one into the other in a linear process (from ‘data’ to ‘wisdom’). Hence, meaningless facts are transformed into knowledge that can be used purposefully.
Figure 4. LA continuous improvement cycle. Adapted from Elias (2011).
Against this background, Elias’s model also consists of a cycle: • Data gathering: capture and select
• Information processing: aggregate and predict • Knowledge application: use and refine (and share)
This process occurs as a combination of tools, actors, theories and organizations (see Figure 4). Elias points out the importance of people, especially when it comes to the implementation of theories and decision-making. Also, the characteristics of the participating organizations, their willingness to support the decisions and actions, and their leadership style play a crucial role (Elias 2011). So, what is the impact of the LA cycle?
Data$ gathering$ Information$ processing$ Knowledge$ application$ capture select aggregate predict use refine Organiza -tion Computers People Theory
Verbert et al. (2013) distinguish four stages in their process model (see Figure 5).
Figure 5. LA process. Source, Verbert et al. (2013).
The visualization of data, e.g., in activity streams, tabular overviews, or other visualizations, is supporting the stage of ‘awareness’. In order to understand the data, the ‘(self-)reflection’ stage focuses on users’ questions, which are derived by reflection. In the ‘sensemaking’ stage, users try to find answer to the questions identified in the previous step. The goal, i.e., the impact, of the iteration is to induce new meaning or change behavior if necessary.
Figure 6. The LA cycle. Adapted from Clow (2012).
Clow (2012) proposes the ‘learning analytics cycle’ presented in Figure 6, which he developed based on Kolb’s experiential learning cycle, Schön’s work on reflective practice, and Laurillard’s conversational framework (Kolb 1984; Schön 1984; Laurillard 2001). Clow’s LA cycle starts with learners, followed by the step
Learners$ Data$ Metrics$ Interven. tions$ The Learning Analytics Cycle
of capturing their data. The next step is processing it automatically or manually into metrics. Lastly, these metrics should be used for interventions that have an effect on learners. This intervention can be a dashboard, which shows the metrics directly to a learner, or a tutor, who contacts the learner personally (Clow 2012). Within the work of this thesis, these descriptions have been revised and integrated into a new, more comprehensive model, which serves as a definition of LA processes with regard to AR. It is presented in chapter 5 and it was useful for evaluating the impact of LA in chapter 7 and verifying requirements in chapter 8.