How data science can be used in elearning?
Posted by Greten on 26 Aug 2019 under Thoughts
Data science is one of the buzzwords that has been circulating in the technology and business communities. The big idea is to extract a large amount of data, organized or not, and process them using computers to see trends and patterns that can guide in decision making. It also includes machine learning: feeding data into computer programs so that they can analyze it, look for patterns, and decide on their next course of action.
At least, that's how I understand it in my slow pace studying of data science. I've been an instructional designer for more than seven years, and I am self-studying to become data scientists as another parallel career. Self-study and slow pace because other real-life responsibilities prevent me from studying full time. Also, I do not have sufficient income to cover for studying it at a university. As I am still learning, any corrections of my current ideas will be much appreciated.
While elearning automates the lesson delivery and allows the learners to work at their own pace and at their own time, data science can aid in other aspects of the learning process such as the automation of spaced learning or repetition, and analysis of the feedback to further improve the existing learning materials.
LMSs have all the structured data data
There are two kinds of data that data scientists worked with, structured data and unstructured data. Structured data are those with clearly defined data types and recorded in a manner that makes it easy for software programs to put them in variables for processing. A list of people with names, college degrees, addresses, and monthly income in a spreadsheet format is an example of structured data.
Unstructured data are those with no apparent data types and can be difficult to process using a software application. An audio recording of people being interviewed about how much they earn and how much they enjoy their work is an example of unstructured data.
Elearning is commonly conducted using a learning management system or LMS. LMSs runs in PHP and stores data such as the quiz scores and progress of learners in SQL database, making them structured data that software can readily process.
Going beyond grade statistics and remedial education
In the traditional education setting, teachers usually look only on the grades. Everyone passed a predefined mark? Good! Some students didn't pass? Well, let us send them to remedial classes where previous lessons are repeated until they passed. The only issue here is the waste of time since students repeat everything, including those topics that they already understand.
Spaced repetition is similar to remedial classes except that they are not necessarily implemented for learners that show unsatisfactory performance; they are used to reinforce the learning for all learners.
With data science, an LMS can have a feature so that spaced repetition can focus on a particular lesson or particular parts of a lesson that the learners are having difficulties. For example, all lessons are repeated in a shorter format at least once for all learners. However, for those learners who have difficulties in a particular topic, the shorter format lesson that covers that topic is repeated twice or more as the remedial lesson. What would the LMS use as a basis of which topics to repeat for a particular student? The LMS can use the quiz results provided that the system stores data of how well the student did in a specific topic, not just the overall quiz.
Thus, an LMS that uses data science can help students to zero-in to a particular topic where they really need a remedial lesson, instead of repeating the entire subject for an entire quarter or an entire year.
Improving the feedback loop for lesson improvement
In a classroom setting, teachers or lecturers provide lecture, write stuff on the board, or show a presentation, and then use quizzes to measure how much the learners learned. Sometimes, a teacher with keen attention to details will notice one or a few multiple-choice items in which many students are selecting the same wrong answer. The teacher could hypothesize that the way he or she is presenting the lesson is prone to erroneous interpretation. Then, the teacher can revisit the lesson plans and adjust his or her teaching strategies accordingly.
In elearning, patterns like this are easier to notice. Furthermore, you can apply data science to run a program that intentionally looks for patterns of consistently wrong answers that might be due to wrong interpretation. Moreover, if your LMS stores information about the learners such as nationality, cultural background, or current city of residence, you can find patterns in which the wrong interpretation of the lesson is due to cultural context and other factors that are valid only to a particular group of people.
Hence, you can take the patterns as feedback to your existing elearning materials, use the feedback to improve the elearning process, deploy again, see if the same or other patterns emerge, and let the loop of feedback and improvement continue until you have learning modules and overall elearning program that work best to as many learners as possible.
Data scientists are the new team members on the block
An elearning team is typically made-up of subject-matter experts (SMEs) to prepare content or ensure its accuracy, instructional designer to design effective lessons from the content, media specialists to come-up with impressive multimedia presentation using the plans from instructional designer, and LMS administrators to curate the modules and ensure that the LMS is working for both learners and the other members of the elearning team.
Data scientists are posing to become the team's new members, ready to look for patterns within the vast quantity of data to improve the learning process. The data are readily available within the LMS' database, and it will be a waste not to utilize them.
The potential to apply data science to elearning is vast and not limited to what I just mentioned. What are the other applications or possible applications of data science to elearning? Let us know in the comments section.
Bibliography:
- Casebourne I. (2015) "Spaced Learning: An Approach to Minimize the Forgetting Curve", Association for Talent Development, retrieved 25 April 2019
- Kim J. (2014) "Here Come the Data Scientists", Inside Higher Ed, retrieved 25 April 2019
- Kozyrkov C. (n.d.) "Data scientist: The sexiest job of the 22nd century", Towards Data Science, retrieved 25 April 2019
- Taylor C. (2018) "Structured vs. Unstructured Data", Datamation, retrieved 25 April 2019
Last updated on 26 Aug 2019.
Share your thoughts
* Required. Your email will never be displayed in public.