Stealing pages from the server...

Difference between Marginal Distribution and Conditional Distribution


Introduction

The possibility of two events occurring at the same time is known as joint probability. To better know the concept, we should take a closer look at marginal distribution and conditional distribution into detail.

Main Difference

  • Marginal Distribution: A marginal distribution considers a subset of variables. (What distribution would I get for if I ignored ?)
  • Conditional Distribution: A conditional distribution fixes a subset of variables. (What distribution would I get for if I set ?)

To summarise what this is attempting to say, conditional distributions are concerned with determining probability for certain subsets of the population under consideration. They deal with determining the probability of one random variable given certain restrictions on the second random variable given two jointly distributed random variables. To put it in other words, a marginal distribution is the distribution of one random variable without any reference to the second random variable.

Marginal Distribution

For , we have marginals and . For discrete values, sum out the others:

For continuous values, integrate out the others:

Conditional Distribution

For both discrete and continuous, divide joint by marginal. The “given” is denoted using the pipe “|”” operator.

For continuous values, integrate out the others:

Real-life Example

In a sample space with more than two events, if only the probability of a particular event occurring individually is considered, it is called the marginal probability. For example,

  • : person who is from UoS
  • : person who is from NCHU
  • : person who is male
  • : person who is female
Male Female Sum
UoS
NCHU
Sum

where , , , are called marginal probabilities. As for conditional probability, for example, given the person is from NCHU, the probability of this person is a female is .

Product Rule and Bayes’ Rule

  • Product Rule:

In probability theory, product rule also called chain rule, which allows you to calculate any member of a collection of random variables’ joint distribution using just conditional probabilities. The rule is useful in the study of Bayesian networks, which are probabilistic networks that characterise a probability distribution using conditional probabilities.

  • Bayes’ Rule

In probability theory and statistics, Bayes’ rule describes the probability of an event, based on prior knowledge of conditions that might be related to the event.

Conclusions

The marginal probability is the probability of a single event occurring, independent of other events. A conditional probability, on the other hand, is the probability that an event occurs given that another specific event has already occurred. This means that the calculation for one variable is dependent on another variable.

References

  1. https://www.youtube.com/watch?v=CQS4xxz-2s4
  2. https://en.wikipedia.org/wiki/Marginal_distribution
  3. https://murphymind.blogspot.com/2011/10/probability.html

Author: Yang Wang
Reprint policy: All articles in this blog are used except for special statements CC BY 4.0 reprint polocy. If reproduced, please indicate source Yang Wang !
 Previous
Batch Normalisation and Layer Normalisation Batch Normalisation and Layer Normalisation
The standardisation of inputs may be applied to input variables for the first hidden layer or to the activations from a hidden layer for deeper layers. In common, this normalisation technique is used on the inputs to the layer before or after the activation function in the previous layer. Using normalisation technique, in addition, can make the network more stable during training. In this articale, batch normalisation and layer normalisation will be compared.
2021-09-17
Next 
One-way ANOVA Implemented from Scratch One-way ANOVA Implemented from Scratch
ANOVA (ANalysis Of VAriance) test used to compare the means of more than 2 groups, whilst t-test can only be used to compare 2 groups. ANOVA uses variance-based F test to check the group mean equality. There are two main types of ANOVA, one-way (one factor) and two-way (two factors) ANOVA.
2021-08-29
  TOC