Skip to main content icon/video/no-internet

Educational Data Mining

Learners leave behind many traces when interacting with an electronic learning environment. Students who use an online environment leave behind their interactions with, for example, simulations, the electronic products they produce (such as concept maps and models), and their chats. Students who use a learning management system (LMS) also leave traces, such as the duration of their visit, the documents they downloaded, the pages they visited, and so forth. Analyzing these data with the goal of discovering unexpected relational patterns is the subject of educational data mining; in the related field of learning analytics, learners’ online data are examined in order to detect predefined patterns. The distinction between the two fields thus basically concerns pattern searching versus pattern recognition. Educational data mining and learning analytics are rapidly growing fields in educational research and practice, driven by the need to adapt learning environments to group or individual needs. This entry first discusses data sources and analysis techniques for educational data mining and learning analytics before discussing how these practices are used.

Data Sources and Analysis Techniques

Learning analytics can be applied to data from a number of sources through a number of techniques and can be used for a number of goals. The most often used data sources are log files of student actions. Every action taken by a student can be registered and used for analysis. This analysis may concern students’ interactions with specific applications such as online laboratories or simulations, or usage of more general software environments such as an LMS. In the first case, the analysis concerns fine-grained student behavior such as a student’s experimentation strategy, while in the second case, it concerns broader actions such as downloading specific files, completing a portfolio, signing up for a course, or data such as exam scores.

The analysis of activity log files primarily relies upon statistical techniques, but a central concern is that these data have no standard format. Another issue is that no standard techniques are available when learning analytics go beyond the analysis of frequencies to also determine sequential patterns. A second set of sources are products that students create while interacting with the learning environment, such as concept maps, models, or portfolios. The automatic analysis of these products may require techniques that are very specific for the type of product, for example, to determine the relatedness of concepts within a concept map or to compare a student concept map to an expert one, or more general techniques involving natural language analysis to determine content-related aspects of a portfolio. A third source of information comes from students’ social interactions and concerns, their interaction patterns, and the content of chats. Network analysis and natural language processing are the relevant techniques here.

Outputs of Results

Results from educational data mining and learning analytics can inform learners, teachers, or applications. Learners and teachers can view the (visual) presentation of information on what are called dashboards. For example, these dashboards can display page visits, downloads, uploads, visits per time unit, and the like, as group statistics or as subgroups or individuals against a group average. Or they can inform students how much of a model they have created is correct. Teachers can use such information to adapt their courses, and learners can utilize it to adapt their learning behavior. For example, students who see that they have not downloaded information already downloaded by the rest of the class may begin to do so. Learning environments can adapt what they offer to students; perhaps, assignments students receive may depend on their scores on previous assignments or students may get automatic advice when their experimentation behavior in a simulation is not systematic. An important aspect of learning analytics is whether they are performed post hoc or concurrently, in which case immediate display or adaptation is possible.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading