By Dr Srabashi Basu, Professor and Program Director, Great Learning
It is commonly believed that education, at any level, is an individual journey. For some, mathematics comes as easy as breathing while others immerse themselves in literature. The same assessments pose differentiated difficulty to different learners, even when they have been exposed to identical learning material and teaching tools. Precision Education is a data-driven approach to tailor teaching and learning to individual requirement, strength, and weakness, and learning style of each student. The idea of precision education originated at the turn of the millennium but the term gained formal recognition in 2016 as technologies advanced. Artificial intelligence and big data analytics, among others, allow educators to customise content, pace, and support for each individual learner. Precision education shies away from the traditional one-size-fits-all model by identifying patterns in student behaviour, performance, and preferences.
The goal is to optimise learning outcomes by addressing specific challenges and leveraging individual strengths, creating a more effective and engaging educational experience for every student. As educators in a blended environment, we focus on the average students. U.S Department of Education recognised that learning objectives, instructional approaches, and instructional content (and its sequencing) may all vary based on learner needs. If applied effectively, learning activities become more meaningful and relevant to learners. Self-motivation is the best pathway to success. Proper assessment identifies the triggers for motivation.
Predictive analytics in education: Identifying at-risk learners
Multiple studies have been done in this area. The primary research objective of using a machine learning approach for precision education happens to be prediction of learning performance, such as drop-out, attrition or retention. Along with that the educationists have worked on student profiling and early identification of at-risk learners. The most frequent learning environment considered is online and data is collected from students’ log files. STEM and Computer Science related learning domains are the most popular domains of study. Research questions on emotional engagement, decision quality, behaviour intentions and brain organisation have also being addressed. An artificial neural network algorithm named Self Organising Maps have been developed which is able to successfully group heterogeneous struggling children into multiple clusters with distinctive cognitive profiles for differentiated intervention. Simpler algorithms, such as Support Vector Machine, has been applied to examine emotional engagement differences in game-based and non-game-based learning.
To develop data science models behavioural characteristics, such as activity log, time and effort devoted to the learning activities, and affective factors have been investigated rather thoroughly. The affective factor includes psychological traits like motivation and attitude, as well as learning styles, learning strategies, coping strategies, relationships with teachers and peers and time management. Based on the VARK (Visual, Auditory, Read/Write and Kinaesthetic) learning style, individuals are classified as Introverts or Extraverts. Introverts learn better by themselves in a quiet place whereas extroverts perform better working in a group or using an audiobook or video, perhaps aided by a background music. Various training models included demographic information, academic characteristics (past performance or prior knowledge) and cognitive factors (reasoning or working memory) as independent variables. Algorithms like K-nearest neighbours, Naïve Bayes, regression, random forest, neural networks, decision tree and SVM have been used most often for research in precision education.
Harnessing algorithms for personalised learning interventions
Personalised learning in the traditional setting is individual tutored learning, which is resource heavy and time consuming. Human interaction factor may introduce negative bias even in the setting. With advanced technology, data-driven precision education is an achievable education paradigm. Rapid advances in automatic emotion detection techniques, eye movement and other body language indicators may open up new possibilities to monitor students’ real-time engagement in learning.
Optimising learning outcomes with data-driven feedback
Providing immediate individualised feedback and intervention is also not impossible. A well validated system can predict which methods or resources work best for each student, adjusting the curriculum, pace, and learning activities in real-time to optimise engagement and outcomes. Learning analytics can track student engagement, participation, and performance in granular detail through video monitoring in classrooms. This feedback can be used by both students and educators to adjust learning approaches quickly. Dashboards may provide real-time insights into student progress and areas requiring attention.
In essence, data science will not only enhance personalisation but also empower educators and institutions with actionable insights to continuously improve learning outcomes, creating a more responsive and inclusive educational system.