Program predictions

- Could you imagine life without preventative medicine? Every day lives are saved because medical experience and research gives us an idea of what to expect in the future without intervention. It's not an entirely new concept, but that doesn't mean it's easy to put to use. A lot of knowledge goes into knowing what to prescribe no matter the situation. Predictive program analytics uses information about the performance of past learning programs to make recommended improvements to future learning programs. To do so, we want to start by identifying the strongest correlations between inputs and resources to the learning program and successful outcomes for the program that we've defined. This helps us evaluate why something was successful so that can inform future predictions or prescriptions. In our case study, we're training Souder Manufacturing managers to take advantage of newly implemented enterprise planning software. Here the participants have a variety of learning options, including elearning courses, on the job coaching sessions, collaborative learning, and a testing environment of the enterprise planning software to explore. So now we want to understand what enables someone to effectively use and navigate this enterprise planning software. We can hypothesize that practice makes perfect. People who spend the most time in the test system have the lowest number of errors in the live system. To test this hypothesis, we review their performance on the related elearning modules and what they've done in the testing environment. Additionally, we capture time spent using the test system and error logs from the live system. We next run a regression analysis to determine the relationship between the data points collected. If our hypothesis that the practice improves the learner's outcomes it correct, we can confidently predict that spending more time in the test system leads to better outcomes in the real environment. Equipped with this information, let's think of ways to incentivize spending additional time in the test system. For example, if someone begins to lag behind their peers in the assessments, the answer might not be additional course work. Instead, allowing for more time to practice and comprehend the concepts may be more beneficial. Exploring new ways to improve outcomes is an application of prescriptive analytics. One of the other goals of Souder's managerial education program was to improve the capability of managers to build strong teams. Using a similar strategy to the previous example, we can test the hypothesis that more frequent on the job coaching leads to teams with less friction and turnover. If that proves true, we can use that information to prescribe additional coaching activity to increase or maintain high levels of teamwork outside of the training program. If you could prove that a specific learning input or resource from your last learning program can make a positive outcome 10% more likely, then you can make a plan for how you can take prescriptive action the next time to increase the overall impact of your learning.