When a model seems satisfactory based on performance, fields included, and the relationships between the predictors and the target, the next step is model assessment. Formally, model evaluation is the assessment of how a model performs on unseen data. Modeler makes this easy because of the Partition field.
We previously used the Partition node to split the data file into Testing and Training partitions. In the previous chapter, we were careful when studying the model not to use the Testing partition. Doing so would compromise model testing because we would learn how well the model performed on the unseen data. In this chapter, we will use the Analysis and Evaluation nodes to further assess our model. ...
