Data analytics in higher education

Data analytics in higher education have gained the attention of Huffington Post and are discussed in a recent article detailing the advancements in how data can be used in universities. The data can be useful in predicting whether students are in trouble academically so their problems can be dealt with more immediately. The article quotes, “Predictive analytics hold the promise that faculty and administrators can intervene at an earlier juncture and help positively impact learning outcomes that ultimately lead to higher graduation rates.”
The article states that the difficulity in managing data is retrieving it from different sources and compiling it to make it into a usable format. The data comes from sources such as individual professors who make the grades, and from a university database that holds information on students such as classes they are enrolled in and whether they have failed a class or not. The most important aspect of predictive analysis, in my opinion, should be managed at the individual professor level. The university is unlikely to have the ability to predict whether a student will fail a class based on a general schema for all classes. Some classes are based on only two exams and a small percentage for homework. If a student failed the first homework, for example, the prediction for this student be that he or she will fail the class. The failure of the homework may or may not reflect on how well the student will do on the exam, which makes up a significant portion of the grade. The predictive analysis will have little accuracy in this case, unless it is done at the individual professor level. The professor would be able to tell much better for the way his or her class was taught if a student was likely to fail. Predictive analytics may not be the best answer in identifying and helping struggling students.
While I disagreed with some aspects of the article, I found that it did contain some interesting links. One link was to Purdue University’s Course Signals. This resource for students “detects early warning signs and provides intervention to students who may not be performing to the best of their abilities before they reach a critical point”. It relies on individual instructors to judge the criteria for an at-risk student, alert them, and provide them with resources for improving their grade in the class.
This article focuses on the ability for data analysis to increase graduation rates and quickly predict whether a student is struggling. I do not think that available data would necessarily provide for an accurate estimation on if a student will fail a class, since many variables influence the success of a student. A student may struggle at the beginning of a class because he or she did poorly on a single assignment, or may not have the same educational grounding as other students. These factors do not predict accurately if a student will fail a class because the learning process occurs over an entire semester. Success cannot be judged within a small enough time period that a data analysis tool would be useful in predicting failure or success as accurately as students and instructors who are directly involved with the class. The predictive analysis must be done at an individual class or even student level, and therefore does not rely on data nearly as much as the article states.
Link to article:<;

This entry was posted in Uncategorized. Bookmark the permalink.

3 Responses to Data analytics in higher education

  1. mfrohman says:

    The title of this post caught my eye as that is exactly what I did this past summer: data analytics in higher education. I interned in the analytics department of a start-up company in higher education. The company that recently changed their name from 2tor to 2U allows students anywhere in the world to earn their master’s degrees online. Three of the schools and programs that 2U is currently partnered with include University of North Carolina, Chapel Hill’s MBA, Georgetown University’s Nursing, and Washington University’s LLM. As an analytics intern, part of my tasks included predictive modeling, just as you described in the Huffington Post article. I took data of the acceptance rates of students into each of the programs and clustered students by demographics and acceptance rates. From there, I then grouped recent applicants in the same clusters and predicted whether they would be accepted into the programs are not.

    This predictive modeling helped to market differently to the various target audiences and other related tasks. With technology today, predictive modeling could very well end up being accurate, depending on the data it is used on. For instance, the predictive modeling of the data described in your post did not appear to be consistent, but I believe the work I did this past summer was fairly accurate.

  2. kristenmayer says:

    If this form of data analytics could be fine tuned to accurately predict when a student needs additional help in a class it would be a great resource for universities. A lot of times, students don’t realize they need help, or aren’t proactive about seeking it until too late. With the help of a program that alerts students and/or professors of student difficulty with material early on, more students may be motivated to seek additional help from the professor, tutors, study groups, etc. Overall, this would be beneficial to students, who would likely be able to learn more, and have a better experience with their classes.

    • mattpuls says:

      The fine tuning of data analytics would be the main issue for a tool like this. Each class has specific grading weights and styles, and there is a lot of variance in the times that grades are obtained. Some classes focus on midterms and less on homework assignments, while others weigh homework assignments heavily and have other graded work that determines more of the grade. An instructor would be better off making their own threshold to determine if a student really needed help than trying to make a system that applies to all classes. A single program would need to be extremely flexible in order to be able to provide good insight as to whether a student simply fell behind for an assignment or was in danger of failing a class to the point that it would be better off to have the instructors do it on a course-level rather than at a university-wide level

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s