About supervised machine learning, in the context of Linear Regression; Decision tree; Logistics Regresion; Randon forest and neural network, which metrics should be used to know if a classification model is good and why use them? What about regression, what metrics are used to say how good the model is? What is the best for each case?