Learning Analytics

Learning Analytics Framework

The result of the increased availability of vast administrative systems with a great deal of academic and personal information is that educational data management, analysis, and interpretation are becoming complex. Several concepts closely linked to processing such educational information are educational data mining, academic analytics, and learning analytics. However, these concepts are often confused and lack universally agreed upon and applied definitions. Educational data mining (EDM) refers to the process of extracting useful information out of a large collection of complex educational data sets. Academic analytics (AA) is the identification of meaningful patterns in educational data to inform academic issues (e.g., retention, success rates) and produce actionable strategies (e.g., budgeting, human resources). Learning analytics (LA) emphasizes insights and responses to real-time learning processes based on educational information from digital learning environments, administrative systems, and social platforms. Such dynamic educational information is used for realtime interpretation, modeling, prediction, and optimization of learning processes, learning environments, and educational decision making.

Figure 1 illustrates a holistic LA framework, linking various types of educational information in a meaningful way.

Information about the learners’ individual characteristics (shown as 1 in Figure 1) includes sociodemographic information, personal preferences and interests, responses to standardized inventories (e.g., learning strategies, achievement motivation, personality), skills and competencies (e.g., computer literacy), prior knowledge and academic performance, and institutional transcript data [Page 448](e.g., pass rates, enrollments, dropouts status, special needs). Information from the social Web (2) includes preferences of social media tools (e.g., Twitter, Facebook, LinkedIn) and social network activities (e.g., linked resources, friendships, peer groups, Web identity). Physical data (3) include information about the learner’s location, sensor data (e.g., movement), affective states (e.g., motivation, emotion), and current conditions (e.g., health, stress, commitments).

Figure 1 Holistic learning analytics framework

Rich information is available from learners’ activities in the online learning environment (4) (i.e., learning management system, personal learning environment, learning blog). These mostly numeric data refer to logging on and off, viewing or posting discussions, navigation patterns, learning paths, content retrieval (i.e., learner-produced data trails), results on assessment tasks, responses to ratings and surveys. More importantly, rich semantic and context-specific information is available from discussion forums as well as from complex learning tasks (e.g., written essays, wikis, blogs). Additionally, interactions of facilitators with students and the online learning environment (OLE) are tracked.

Closely linked to the information available from the OLE is the curriculum information (5), which includes metadata of the OLE. These data reflect the learning design (e.g., sequencing of materials, tasks, and assessments), and learning objectives as well as expected learning outcomes. Formative and summative evaluation data are directly linked to specific curricula, facilitators, or student cohorts.

The LA engine (6) is based on pedagogical theories and methodological and mathematical approaches. Rich information from various sources (i.e., structured and unstructured data) is processed using specific algorithms (e.g., Bayesian networks, neural networks, natural language processing, survival analysis, hierarchical linear modeling) that are closely linked to the underpinnings of applied pedagogical theories. The results of the data mining process are validated before further analysis is computed for real-time comparisons and identification of patterns, as well as for predictive modeling. The reporting engine (7) uses the results of the LA engine and automatically produces useful information in the form of interactive dashboards, heatmaps, statistics, graphs, and automated reports. These automated reports are used for specific stakeholders such as the governance level (10; e.g., for cross-institutional comparisons), single institutions (9; e.g., for internal comparisons, optimization of sequence of operations), and the curriculum level (4), including insights and reports for learning designers and facilitators for analyzing instructional processes and students’ pathways.

...