Understanding how far an educational institution has gone on the implementation of learning analytics, may be achieved by using an “analytics maturity model”. There several models available.
One of the most significant models comes from Gartner. Garner describes his analytics maturity model by multiple dimensions:
- Time perspective covered: past, present or future
- Human input required to support decision and action processes
- Data analysis mathematical complexity
The following diagram depicts Garner’s model:
The least advanced level comprises descriptive analytics. Those analytics provide information about the past but in terms of pure statistical data.
Average grade in a course or percentage of students reaching the end are typical examples of analytics falling within this category.
Further human understanding is required to derive why the number of students finalizing the course was not as expected.
Descriptive learning analytics are obtained using basic calculations.
On top, we have diagnostic analytics. Although also focused on the past, they try to explain the reasons for specific outcomes. Therefore, they use more complex algorithms to determine causality.
Not all students finalized the course because the course content was to extensive. This is an example of a diagnostic analytic.
Using these analytics, we can influence and improve future instances of our learning processes. Action is closer, but still requires quite a lot of human input.
On a different level are predictive analytics. Those are focused on the future. They anticipate outcomes and enable on time improvement actions.
Dropout rate predictions are a clear example of this type of analytics.
Complex machine learning algorithms are used to create predictive analytics. Having relevant and enough data are major factors influencing the predictions accuracy.
There is an important benefit for those institutions capable of predicting the future. They can timely act on ongoing learning processes mitigating risk and improving performance.
On top of the pyramid we have prescriptive analytics. Using prescriptive analytics, organizations get predictions about the future as well as action recommendations based on those predictions.
These analytics may infer that a given student will not reach his learning objectives and suggest reinforcement content or additional learning activities.
The automatic generation of prescriptions (recommendations) requires even more complex algorithms. Once again, data volume and relevance is critical for success.
Every step towards a higher level of learning analytics maturity is built on top of the previous one.
Organizations capable of moving away from descriptive/diagnostic analytics into predictive/prescriptive analytics will gain an important competitive advantage from the business and educational perspectives.
Look at this analytics maturity model within your organization. Ask yourself whether your institution has already developed these capabilities. Have in mind that this is not a simple process. It requires not only important technical skills but overall organization alignment. The effort is well worth.
By submitting this form, you accept the privacy policy