Introduction

When dealing with jointly distributed random variables, we are sometimes interested in the conditional probability distribution function of one random variable given the that the other random variable takes on a particular value.

Conditional Distribution

Let \(X\) and \(Y\) be a pair of jointly distributed random variables. The conditional probability function of the random variable \(Y\), given that the random variable \(X\) takes the value \(x\), is \[p(y|x) = \frac{p(x,y)}{p(x)} \text{ for all values of } y\] Similarly, the conditional probability function of the random variable \(X\), given that the random variable \(Y\) takes the value \(y\) is \[p(x|y) = \frac{p(x,y)}{p(y)} \text{ for all values of } x\]

Suppose \(X\) and \(Y\) have the following joint probability distribution

\(x=-3\) \(x=3\)
\(y=1\) \(0.2\) \(0.2\)
\(y=8\) \(0.3\) \(0.3\)

The conditional distribution of \(Y\) given that \(X = 3\) has two components. The first component is when \(Y=1\)

\[p(1|x=3) = \frac{p(1,3)}{p(3)}=\frac{0.2}{(0.2+0.3)}=0.4\] and the second component is when \(Y=8\) \[p(8|x=3) = \frac{p(8,3)}{p(3)}=\frac{0.3}{0.2+0.3}=0.6\] Putting it all together, we get that the conditional distribution of \(Y\) given \(X=3\) is

\(y|x=3\) \(1\) \(8\)
\(p(y|x=3)\) \(0.4\) \(0.6\)

Example 1:

Consider the joint probability distribution

\(x=1\) \(x=2\)
\(y=0\) \(0.30\) \(0.20\)
\(y=1\) \(0.25\) \(0.25\)

Find the conditional distribution of \(X\) given that \(Y=0\).

\[p(x=1|y=0) = \frac{P(x=1,y=0)}{P(Y=0)}=\frac{.3}{.3+.2}=0.6\] \[p(x=2|y=0) = \frac{P(x=2,y=0)}{P(Y=0)}=\frac{.2}{.3+.2}=0.4\] Which gives us

\(x\) \(1\) \(2\)
\(p(x|y=0)\) \(0.6\) \(0.4\)

The Conditional Mean

\[\mu_{y|x} = E(Y|X=x) = \sum_y yp(y|x)=\sum_y y \frac{p(x,y)}{p(x)}\]

Continuing with our example from above, we can use the conditional distribution of \(Y\) given \(X = 3\) to find the conditional mean of \(Y\) given \(X=3\)

\[\mu_{y|x} = 1(0.40) + 8(0.60) = 5.2\]

Example 2:

Consider the joint probability distribution

\(x=1\) \(x=2\)
\(y=0\) \(0.30\) \(0.20\)
\(y=1\) \(0.25\) \(0.25\)

Find the conditional mean of \(X\) given that \(Y=0\).

The conditional distribution of \(X\) given that \(Y=0\) is (from Example 1)

\(x\) \(1\) \(2\)
\(p(x|y=0)\) \(0.6\) \(0.4\)

So the mean is \[E(X|Y=0) = 1(.6) + 2(.4) = 1.4\]

The Conditional Variance

\[\sigma_{y|x}^2 = E[(Y-\mu_{y|x})^2|X=x] = \sum_y (y-\mu_{y|x})^2 p(y|x)=\sum_y (y-\mu_{y|x})^2 \frac{p(x,y)}{p(x)}\]

Continuing with our example from above, we can use the conditional distribution of \(Y\) given \(X = 3\) to find the conditional variance of \(Y\) given that \(X=3\)

\[\sigma_{y|x}^2 = (1-5.2)^2 \cdot 0.40 + (8-5.2)^2 \cdot 0.60 = 11.76\]

Example 3:

Consider the joint probability distribution

\(x=1\) \(x=2\)
\(y=0\) \(0.30\) \(0.20\)
\(y=1\) \(0.25\) \(0.25\)

Find the conditional variance of \(X\) given that \(Y=0\).

From Examples 1 and 2, we have the conditional distribution and the conditional mean. Using these we can find the conditional variance:

\[\sigma_{x|y}^2 = (1-1.4)^2 \cdot 0.6 + (2-1.4)^2 \cdot 0.4 = 0.24\]

The Conditional Standard Deviation

We can just take the square root of the conditional variance to get the conditional standard deviation

\[\sigma_{y|x} = \sqrt{\sigma_{y|x}^2}\]

Example 4:

Consider the joint probability distribution

\(x=1\) \(x=2\)
\(y=0\) \(0.30\) \(0.20\)
\(y=1\) \(0.25\) \(0.25\)

Find the conditional standard deviation of \(X\) given that \(Y=0\).

From Example 3, we have that the conditional variance is \(0.28\). Taking the square root gives our result \[\sigma_{x|y} = \sqrt{0.24}\]

The Continuous Case

The same concepts hold for continuous random variables with integrals replacing the sums. More details on this case are beyond the scope of the course.