Using analytics for learning and development is quickly gaining traction in usage and popularity. This is especially true for large and enterprise organizations. Having the ability to gain data and meaningful insights from learning activities helps organizations improve not only their L&D strategies, but also business performance metrics as well.
Learning analytics allow companies to baseline training performance and also get a view into the past (what has already happened), current (what’s happening now), and even future performance (what is going to happen). This allows organizations to effectively tweak training programs and plan for future initiatives.
The Meshing of Learning and Big Data
Big data exists everywhere. If you use any kind of digital device or platform, there is information being gathered about many variables of you and your life. Aggregated throughout many years, all of this data may even know more about you than some of your closest friends! It is a little unnerving to think about, but if we imagine this concept in the form of learning, we will see that we can use the power of data for the good of the individual learners, managers, and organization as a whole. This isn’t an article to address and discuss GDPR and privacy principles (which is an entire topic in itself), but instead, to uncover the power of analytics when it comes to the learning function.
The world of big data has existed for years and the learning function is finally being brought up to the big data table. This opens a world of opportunity for L&D to begin making data-driven decisions and effectively proving the business case for learning. This isn’t to say that instructional designers will now need to be data scientists, but they will need to have a basic understanding of analytics and be able to tell a data-driven story around their learning programs. However, there is currently, a knowledge gap that exists on the benefits of learning analytics for the organization.
Overall, the purpose of this article is to introduce 3 main types of analytical models that are most commonly used to give you (the L&D practitioner) a better understanding of what companies mean when they refer to learning analytics and big data concepts.
Descriptive Analytics – Understanding the Past
Descriptive analytics is the most basic type of analytical model which covers historical trends (what has already happened). These historical trends can be used to get insight into a population’s behavior and are typically used day-to-day by organizations. Descriptive analytics are great because they are fairly common and quite simple to begin using as well.
Metrics such as average daily views and impressions are examples of descriptive analytics.
The possibilities of descriptive analytics are endless due to their simplicity. Examples of descriptive analytics may be as simple as how many learners took your course, what the grades on the quiz were, how many social shares your resource received on social media, stock inventory, last year’s sales, etc. These metrics are basic in that many can be found within your LMS or easily accessed through other functions of the business (sales, operations, accounting, etc).
Predictive Analytics – Predict What Might Happen
Predictive analytics focuses on probabilities of future situations occurring. It answers the question “What might happen?” through statistical models. Predictive analytics are the next step up from descriptive analytics. We use the historical data that we have gathered as a foundation for the next step in decision making. This often involves data from multiple sources, which can be gathered using a Learning Record Store (LRS). Looking into predictive analytics helps inform you, as well as decision makers, on an event’s likelihood to occur in the future. For L&D, the possibilities are exciting. This may mean being able to predict whether a workplace accident is likely to occur, how long an employee is expected to stay on the job, and more.
Typically, L&D may need to pair up with a data scientist group to do these kinds of correlations and predictions as these involve advanced statistical models and math concepts.
It’s important to note that predictive analytics will focus on predicting a single learner or customer behavior, while descriptive analytics typically describes a population. For example, L&D can use predictive analytics to determine the likelihood of a particular student receiving an “A” on the next exam and use this information to improve course and studying material. Predictive analytics can also be used to determine how likely it is that a learner will apply concepts they learned while on the job which can be used to help guide the design and delivery of future learning materials.
Prescriptive Analytics – Prescribe Solutions for Various Outcomes
Prescriptive analytics is where the fun begins (not that descriptive and predictive analytics aren’t fun!), but prescribing solutions and options is the most sophisticated analytics type of the 3. This involves exploring a set of potential options to recommend the best course of action for a particular situation. These analytics are exciting in that they advise what should be done in the predicted future. The power of prescriptive analytics is in the capability to quantify the effect of future actions on business key performance indicators (KPIs) all while suggesting the most optimal action.
Prescriptive analytics can be a form of machine learning in that these models automatically “learn” from current data to improve intelligence over time. This involves the use of big data and complicated algorithms that compare outcomes of numerous data sets and then choose the optimal course of action to support business objectives. Typically, only large and enterprise companies use this type of modeling, but can have pronounced effects on company effectiveness and decision-making tact.
All types of industries use prescriptive models to inform business decisions. You might even decide to use prescriptive analytics within your learning function to recommend relevant learning resources to individual learners based on a machine learning algorithm.
Thinking to the Future
So, all of this might sound great, but how do we even capture the data to start using these analytical models? It can be done through a variety of ways, but we recommend adopting an interoperable data specification such as xAPI. This allows you to get data from anything digital in a single data format, making the analytics piece much easier for you and/or your data scientist team. If you’re ready to start learning about how to gain meaningful insight into your learning programs, contact Riptide to learn more about Storepoints Learning Record Store (LRS).