Machine Learning for Industrial Applications

9Model Selection

Model selection is the process of selecting the final learning machine model among a set of training data model training models.

Version choice is a technique that can be implemented to each exclusive forms of fashions [e.g., retrieval, support vector machine (SVM), and k-nearest neighbor (KNN)] and all models of the same type configured by different model parameters (e.g., one of a kind kernels in SVM).

If we have a variety of models of different complexities (e.g., line or regression models with different polynomials, or KNN separators with different K values), how should we choose the right one?

For example, we may have a database that has an interest in creating a segment or model that predicts retrospective. We do not know in advance which model will perform better on this issue, as it is unknown. Therefore, we measure and evaluate the list of different models in the problem.

Selecting models is the process of selecting one of the models as the final model that deals with the problem. Selecting models is different from model testing.

For example, we test or evaluate candidate models to choose the best one, and this is a choice of models. Although once the model is selected, it can be tested to determine how well it is expected to work normally; this is a model test.

The process of evaluating the performance of a model is known as model testing, whereas the process of selecting the appropriate level of model flexibility is known as model choice.

9.1 Model ...