In this basic-level chapter, we discussed the basics of learning algorithms and their purpose. Then, we studied the most basic way of measuring success and failure through performance analysis using accuracies, errors, and other statistical devices. We also studied the problem of overfitting and the super important concept of generalization, which is its counterpart. Then, we discussed the art behind the proper selection of hyperparameters and strategies for their automated search.
After reading this chapter, you are now able to explain the technical differences between classification and regression and how to calculate different performance metrics, such as ACC, BER, MSE, and others, as appropriate for different tasks. Now, you are capable of detecting overfitting by using train, validation, and test datasets under cross-validation strategies, you can experiment with and observe the effects of altering the hyperparameters of a learning model. You are also ready to think critically...
