5.4. Assessment of modelsΒΆ
Assess the model has focused on interprets the model based on domain knowledge of the subject analyzed, project success criteria, desired test design, and the data mining results in the business context.
There are several keys steps to develop the model assessment process:
apply a single technique more than once or generates data mining results with several different techniques
compare all results according to the evaluation criteria
summarize results of this task
list qualities of generated models
rank models quality by comparing them with each other
revise parameters settings, tune them and build a new model with it
iterate model building and assessment until found the best model considering the criteria
document all revisions and assessments