How does this work technically?
Just like supervised learning algorithms, prescriptive analytics needs training data to learn insights and connections and then process them into a model. With prescriptive analytics the data is analyzed in a fundamentally different way than with classification, which focuses on predicting a value or event. In the background, Trendskout uses different types of algorithms such as neural networks and gradient descent trees combined with propensity modeling. As an evaluation criterion during the Auto ML & Hypertuning step, historical data is used to evaluate the accuracy of the prescriptions.
After the training phase, and associated hypertuning, the winning model is used to generate the prescriptions. For example the following designated sales actions for an account manager.