Predictive Analytics

Predictive analytics uses existing data to build models that estimate future outcomes, classify observations, or find patterns that help explain what is likely to happen next. In the statistics calculator, this section groups methods that are used for prediction, classification, and segmentation.

The available predictive analytics methods are currently the Decision Tree calculator and the Random Forest calculator and the Gradient Boosting calculator. Decision Tree, Random Forest, and Gradient Boosting automatically run classification for categorical targets and regression for metric targets. CHAID remains available as an additional decision tree method for categorical targets.

Predictive analytics methods

Predictive analytics methods are especially useful when the goal is not only to test a single hypothesis, but to create a model from several variables. Depending on the method, the result can support classification, customer segmentation, risk scoring, or decision support.

More predictive analytics options can be added to this section while the calculator keeps the same workflow: select a method under Calculate, then use the How-to section for method-specific explanations.

Przykłady z zarządzania jakością, takie jak przewidywanie braków, ryzyko reklamacji i analiza parametrów procesu, znajdziesz w samouczku Analityka predykcyjna dla danych jakościowych.