Marginalization is a method of discovering one variable based on another variable. It determines the marginal contribution of another variable by summing up the possible values of the variable. This definition is relatively abstract, and it is described below using relevant cases.

Assuming that we need to understand the impact of weather on happiness, we can use P (happiness | weather) to express it, that is, given the weather type, what is the approximate level of a person's happiness index.

Marginalization tells us that we can get the desired probability quantity by simply adding up some probabilities, and if we calculate the answer (including single values or distributions), we can get any desired properties. In addition, integrating or summing variables can also be called marginalization.

References:

Concept Analysis of Probability Theory: Marginalization(Marginalisation)