FAQ 4: What is the difference between predictive and prescriptive analytics?
In this series you will find answers to questions we get asked a lot when speaking about data, analytics, digital transformation, emissions reduction, industry 4.0, IoT and data-driven automation.
What is the difference between predictive and prescriptive analytics?
To achieve end-to-end automation with analytics we need more than predictions. It is not enough to know that a machine will break or the yield will be off-spec. We want to know when will this happen, and what should we do to prevent this from happening. Prescriptive analytics provides recommendations on corrective actions we can take to improve the outcomes. It is actionable, as it lets models work for us continuously until the process is changed so much that input-output relationships change. If this happens, the models should alarm us with a large error between predicted outcomes and observed outcomes. At this moment we would re-train the models, discover new input-output relationships, validate them and deploy again to give real-time corrective feedback.