17.1 Related concepts
Logistic regression is more about modeling from a probability level, but in practical applications, it is more used in binary classification problems.
Thinking: How to calculate "better fitting", analogous to the least squares method of linear regression.
sigmoid function: