In predictive analytics, the aim is to build an analytical model predicting a target measure of interest.1 The target is then typically used to steer the learning process during an optimization procedure. Two types of predictive analytics can be distinguished: regression and classification. In regression, the target variable is continuous. Popular examples are predicting stock prices, loss given default (LGD), and customer lifetime value (CLV). In classification, the target is categorical. It can be binary (e.g., fraud, churn, credit risk) or multiclass (e.g., predicting credit ratings). Different types of predictive analytics techniques have been suggested in the literature. In what follows, we will discuss a selection of techniques with a particular focus on the practitioner's perspective.
Because the target variable plays an important role in the learning process, it is of key importance that it is appropriately defined. In what follows, we will give some examples.
In a customer attrition setting, churn can be defined in various ways. Active churn implies that the customer stops the relationship with the firm. In a contractual setting (e.g., postpaid telco), this can be easily detected when the customer cancels the contract. In a noncontractual setting (e.g., supermarket), this is less obvious and needs to be operationalized in a specific way. For example, a customer churns if he or she has not purchased any products during ...