One general concept that might need clarification is the distinction between "descriptive" learning analytics, i.e. reports, and "predictive" learning analytics, i.e. models driven by machine learning.
Possibly we should have a separate course about descriptive reporting, because a lot of people seem to have questions and projects based around this toolset. Descriptive reports collect data from a wide range of courses and/or users and summarize the data for interpretation by a skilled human, but they don't make predictions, and any calculations are manually defined by the report designer.
Predictive analytics models, such as those supported by the Moodle Learning Analytics API, take a number of possible indicators (predictors) and a defined goal (target) and use an algorithm to determine how the indicators relate to the target. The model developer doesn't have to determine which indicators should have the most weight or how to calculate the prediction-- machine learning takes care of that based on historical data.
Predictive models require that there is some target or goal that can be agreed upon as a "known good," whereas descriptive models are often used to explore a system to try to determine what data elements might make good predictors or targets.Currently, our courses are all intended to help with the design and development of predictive analytics models using the Moodle Learning Analytics API. Is there demand for a descriptive reporting workshop? I've also posted this question to the Working Group Roadmap Forum. Please respond there with your thoughts. Thanks!