To distinguish the Model Specifications or Parameters from data item specifications (or parameters), the Model Specifications are typically referred to as Hyperparameters. Finding the best ones to use is called Hyperparameter Tuning.
Hyperparameter tuning requires some insight into how the model will be used. This is normally based upon one or more model metrics. For example, in a case of Fraud/No-Fraud or Loyal/Churn one of these classes is typically much smaller than the other (Fraud or Churn). In one application you might want to make sure that the model catches as much Fraud as possible. In another application, you might want to have the model catch only Fraud with the highest confidence level and pass through as many No-Fraud cases as possible. In another, you might want a more balanced approach.