Evaluation Criteria and Model Selection

This chapter demonstrates the tools to evaluate and select appropriate models for machine learning applications. This chapter begins with theoretical explanations and mathematical examples of different error criteria. Then, some frequently used distance metrics in machine learning are briefly explained. Some mathematical examples and Python code implementation are provided for error and distance metrics. Then, the chapter discusses the confusion matrix and its associated scores, such as accuracy, precision, recall, and f1 score. Afterward, the chapter introduces the concepts of model parameters, hyperparameters, and some model optimization techniques. This chapter also discusses bias–variance, model overfitting–underfitting, and how to address associated issues. Lastly, this chapter concludes by discussing some essential model selection methods. By the end of this chapter, the readers will be able to evaluate and select suitable models for their machine learning applications.