HKUST Visual Techniques

The system is designed for educators to observe students' behaviors and performances on different questions. The questions chosen as samples are from a math problem platform focusing on junor/middle students. Users could choose different questions on the right bar of the system. Overview part of the system provides three views, which are ROI view, Heatmap View and Transition View. You can choose view in the Control Panel. And the Control Panel also let user to adjust parameters for each view. The Data Analytics part has two statistic bar charts, where one is the action distribution and another is the score distribution of the chosen question. The Transition Map part is used to show the transition map of every student who has done the chosen problem and is also embedded a button to give a slot. The slot could show the cluster result of all the transtion maps, which may indicate some insights in students' learning behaviors.

MIT code repositories

The MOOC Learner Project taps the potential of Massive Open Online Course student behavioral data by providing data science technology that makes the data accessible for teaching and learning research. It enables insights into how students learn and how instructors can effectively teach. It is developed by [ALFA-Group].

HKU LDSHE (Learning Design Studio for Higher Education)

Learning Design Studio HE (LDSHE) is a pedagogically grounded productivity and collaboration platform for professionals in the Learning Design (LD), Learning Analytics (LA), and Education communities (particularly those interested in using LD to support Teacher Inquiry of Student Learning (TISL)). To do so requires a common design language that can (1) capture well-constructed pedagogical practices and the underpinning learning design principles, as well as specify the necessary learning analytics appropriate for the intended learning outcomes and chosen pedagogy, and (2) be understood by practitioners and researchers in all of the three targeted communities.

Learning Design

Idshe-docker can help MOOC users (instructors, instructional designers, institutional curriculum leaders and learning scientists) to find a good way to better design their MOOC courses.

Data Analytics

We developed a series of data analytics techniques to analyze the MOOC data (e.g., mouse clicks, video controls, problem responses, programming, collaborations and discussions). The goal of our data analytics techniques is to tap into the immense potential of the MOOC data to provide insights into how students learn and how instructors can effectively teach. Our data analytics techniques have been integrated into two architectures. One architecture is Mooc-Learner-Project (MLP), which contains 3 sub-components: Curation, Pipeline and Data Science Analytics. The second architecture is MLDSA (MOOC Learner Data Science and Analytics). MLDSA allows the educational data scientist to apply deep neural networks, Auto Encoders and Transfer Learning to the MOOC data analysis. It enables experimentation with different labels, features, neural architectures and parameters. These data analytics techniques can help users gain quick insights into the MOOC data.

Data Visualization

We developed novel visualization techniques to help domain experts to analyze large-scale data of MOOCs. It provides course instructors and education analysts with intuitive, interactive and detailed analysis of the MOOC data including clickstream data when students interact with course videos, grading data for assignments and exams, and forum data. More specifically, our data visualization techniques help users explore the MOOC data from the following perspectives: video popularity, click stream analysis, demographic distribution, user forum interactions, dropout analysis and sequence analysis. Multi-exploration techniques are offered for analysis at different levels. With the help of our data visualization techniques, course instructors/designers and learning analysts can conduct detailed exploration of the learners’ learning behaviors in an intuitive way, gain deep insights into the students’ learning behaviors and further improve their course designs with evidence-based guidance.