The basic questions in a classification task:
(1) How to generate the features? The feature generation process is always problem dependent, and it concerns the feature generation stage of the design of a classification system that performs a given pattern recognition task;
(2) What is the best number L of features to use? The problem concerns the feature selection stage of the classification system;
(3) Having adopted the appropriate,for the specific task, features , how does one design the classifier? What type of non-linearity must one adopt, and what type of optimizing criterion must be used in order to locate a surface in the right place in the l-dimensional feature space? These questions concern the classifier design stage.
(4) Finally, once the classifier has been designed, how can one assess the performance of the designed classifier? That is, what is the classification error rate? This is the task of the system evaluation stage.
the stages for designing a classification system
What we can find from the figure is that, the stages are interrelated and, depending on the results, one may go back to redesign the earlier stages in order to improve the overall performance.