Posted by Juanjuan Wang
Last September, a friend of mine recommended Coursera to me, an online higher education platform that partners with the top universities in the world to offer courses online for anyone to take, for free. After trying 30 minutues, I quickly fell in love with this easy-to-use platform. Soon after that, I discovers several other similar platforms which offers highly interactive university level courses, such as Udacity and Stanford Venture Lab. Among all those website, I like Coursera the most. I would like to call those website online higher education 2.0, because comparing to the open courses you could find on iTunes U and some 1-tier universities' websites, these recent adventures give a well-rounded learning experience including pre-reading, class discussion, real-time quiz, assignment and test. It provides a whole learning environment, which highly imitate that you tranditionally get from offline classroom.
What is more exciting is how programs like Coursera could leverage the power of big data. Big data, according to wikipedia, is a collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools or traditional data processing applications. Coursera already hit one million students in 196 couturiers world wide last August, which means that every second there are students conducting all kinds of activities on their website. This certainly generates enormous scale of data which we could draw insight from.
According to a post on Fast Company website:
"When a student takes a Coursera class, the system tracks the learner's every move. The typical educational study has 20 students; the big ones have 200. Here we have 20,000, or even 200,000, and it just completely changes your ability to extract meaningful conclusions from the data. We can see every single click: pausing, rewinding, the first and second try on the homework, what they did in between."
For example, while reviewing answers of an assignment of Machine Learning, one of the earliest, most popular courses at Coursera given by one of its co-founder Andrew, Andrew noticed that 2,000 users submitted identical wrong answers. Further analysis revealed that the student errors originated with switching two lines of code in a particular algorithm.
Analyzing big data enables more accurate and insightful observation, which could help improving education delivery little by little. In the student forum for Ng's machine-learning class, they found that a particular post helped a lot of students find the correct response to a question on one of the quizzes. According to the post mentioned above, in the next round of the course, Ng might include that information in an FAQ or address it directly on the video.
Big data could help online education in more way than one. Recently I had a very informative interview with Aaron Schumacher, who has worked as an analyst at the NYC Department of Education and is now with NYU's Data Services as Senior Data Services Specialist. In that conversation, Aaron introduced a lot of useful information and websites on how technology and computer could help improve education, some of them would be included in later posts. One of the website he mentioned is Knewton's adaptive learning, which seems really exciting because it makes a big step further in using big data to make real-time individualized education. According to its website, "Knewton is a technology company that uses data to continuously personalize online learning content for individual students. Knewton analyzes data about the performance of each student and similar students on the platform, as well as the relevance of the educational content, in order to serve up the best activity for each student at a particular moment in time."
In specific, Knewton's model is called "Continuously adaptive learning". "The platform is continuously adaptive, meaning it responds in real time to each student’s activity on the system and adjusts to provide the most relevant content". Previously, a lot of practices in education field could serve as an example of the "black box" theory. Although education and psychology experts conducted many researches on learning mechanisms, there is still great uncertainty in which actually works better, what actually influence the learning outcome and how. This is certainly an area where data could jump in. According to a post on Inside Higher Education named "Pearson and Knewton: Big Data and the Promise of Personalized Learning", "the promise of digital content is, in part, the promise to be able to glean more insight into learning. Being able to provide solid data about student progress -- what they understand, what they don't -- in real time and not just at the beginning or end of the semester, being able to provide remediation aimed at those strengths and weaknesses… these are powerful, powerful offerings."
The above mentioned practices remind me of a core idea Viktor Mayer-Schonberger mentioned in his recent book "Big Data: A Revolution That Will Transform How We Live, Work, and Think", that "within this Big Data era correlation seems far more important than causation. Figuring out what variables are predictive gets you far better and rich results than figuring out which ones are truly causal".
Notwithstanding the rising role big data is playing in improving education in this technology era, there is still some lower hanging fruit, which could be achieved with simpler methods. In our conversation, Aaron provided a very good example: Assistments, which is an an online tutoring program that gives immediate feedback to teachers, students, school administrators, and parents created with.
How it actually works? First of all, the teacher assigns a problem set to the student:
(Note: all the pictures here are taken from the website of Assistments)
Then, students could get quick feedback, teacher gets immediate evaluation on how the student and the class performed like below, and as Aaron mentioned, parents could also access data on how their children perform.
What shows above is not comprehensive big data analysis; Assisstment proves that even simple data analysis and data communication could largely benefit all stakeholders and improve the outcome of education.
In this era of technology, there is certainly more to expect. Let's witness it together and hopefully become part of the revolution sometime in the future.