From the perspective of probability theory and decision theory. Probability models are divided into participation method and nonparametric method. The parameter method is ultimately to make a parameter estimation, generally point estimation and maximum likelihood estimation. Non-parametric methods are mainly kernel method and nearest neighbor method, that is, estimating the probability in a region, which category has the greatest probability, then the classified object belongs to this category.

Decision theory is to use some methods to make decisions on the following steps, such as entropy decision, decision tree and so on.

Nowadays, probability model is widely used because it has a good mathematical description. Decision tree method has its advantages in many aspects, and it is suitable for processing big data in terms of speed.

Further reading: How to buy insurance, which is good, and teach you how to avoid these "pits" of insurance.