Change-of-Variable Technique

Printer-friendly versionPrinter-friendly version

On the last page, we used the distribution function technique in two different examples. In the first example, the transformation of X involved an increasing function, while in the second example, the transformation of X involved a decreasing function. On this page, we'll generalize what we did there first for an increasing function and then for a decreasing function. The generalizations lead to what is called the change-of-variable technique.

Generalization for an Increasing Function

Let X be a continuous random variable with a generic p.d.f. f (x) defined over the support c1 < x < c2. And, let Y = u(X) be a continuous, increasing function of X with inverse function X = v(Y). Here's a picture of what the continuous, increasing function might look like:

increasing function

The blue curve, of course, represents the continuous and increasing function Y = u(X). If you put an x-value, such as c1 and c2, into the function Y = u(X), you get a y-value, such as u(c1) and u(c2). But, because the function is continuous and increasing, an inverse function = v(Y) exists. In that case, if you put a y-value into the function X = v(Y), you get an x-value, such as v(y). 

Okay, now that we have described the scenario, let's derive the distribution function of Y. It is:

\(F_Y(y)=P(Y\leq y)=P(u(X)\leq y)=P(X\leq v(y))=\int_{c_1}^{v(y)} f(x)dx\)

for d1 = u(c1) < y < u(c2) = d2. The first equality holds from the definition of the cumulative distribution function of Y. The second equality holds because Y = u(X). The third equality holds because, as shown in red on the following graph, for the portion of the function for which u(X) ≤ y, it is also true that X ≤ v(Y):

graph

And, the last equality holds from the definition of probability for a continuous random variable X. Now, we just have to take the derivative of FY(y), the cumulative distribution function of Y, to get fY(y), the probability density function of Y.  The Fundamental Theorem of Calculus, in conjunction with the Chain Rule, tells us that the derivative is:

\(f_Y(y)=F'_Y(y)=f_x (v(y))\cdot v'(y)\)

for d1 = u(c1) < y < u(c2) = d2

Generalization for a Decreasing Function

Let X be a continuous random variable with a generic p.d.f. f (x) defined over the support c1 < x < c2. And, let Y = u(X) be a continuous, decreasing function of X with inverse function X = v(Y). Here's a picture of what the continuous, decreasing function might look like:

decreasing Y 

The blue curve, of course, represents the continuous and decreasing function Y = u(X). Again, if you put an x-value, such as c1 and c2, into the function Y = u(X), you get a y-value, such as u(c1) and u(c2). But, because the function is continuous and decreasing, an inverse function v(Y) exists. In that case, if you put a y-value into the function X = v(Y), you get an x-value, such as v(y). 

That said, the distribution function of Y is then:

\(F_Y(y)=P(Y\leq y)=P(u(X)\leq y)=P(X\geq v(y))=1-P(X\leq v(y))=1-\int_{c_1}^{v(y)} f(x)dx\)

for d2 = u(c2) < y < u(c1) = d1. The first equality holds from the definition of the cumulative distribution function of Y. The second equality holds because Y = u(X). The third equality holds because, as shown in red on the following graph, for the portion of the function for which u(X) ≤ y, it is also true that X ≥ v(Y):

pdf y

The fourth equality holds from the rule of complementary events. And, the last equality holds from the definition of probability for a continuous random variable X. Now, we just have to take the derivative of FY(y), the cumulative distribution function of Y, to get fY(y), the probability density function of Y. Again, the Fundamental Theorem of Calculus, in conjunction with the Chain Rule, tells us that the derivative is:

\(f_Y(y)=F'_Y(y)=-f_x (v(y))\cdot v'(y)\)

for d2 = u(c2) < y < u(c1) = d1. You might be alarmed in that it seems that the p.d.f. f(y) is negative, but note that the derivative of v(y) is negative, because X = v(Y) is a decreasing function in Y. Therefore, the two negatives cancel each other out, and therefore make f(y) positive.

Yell Phew! We have now derived what is called the change-of-variable technique first for an increasing function and then for a decreasing function. But, continuous, increasing functions and continuous, decreasing functions, by their one-to-one nature, are both invertible functions. Let's, once and for all, then write the change-of-variable technique for any generic invertible function.


Definition. Let X be a continuous random variable with generic probability density function f (x) defined over the support c1 < x < c2. And, let u(X) be an invertible function of with inverse function v(Y). Then, using the change-of-variable technique, the probability density function of is:

\(f_Y(y)=f_X(v(y))\times |v'(y)|\)

defined over the support u(c1) < y < u(c2).

Having summarized the change-of-variable technique, once and for all, let's revisit an example.

Example

Let's return to our example in which X is a continuous random variable with the following probability density function:

\(f(x)=3x^2\)

for 0 < x < 1. Use the change-of-variable technique to find the probability density function of \(Y=X^2\). 

Solution. Note that the function: 

\(Y=X^2\)

defined over the interval 0 < x < 1 is an invertible function. The inverse function is:

\(x=v(y)=\sqrt{y}=y^{1/2}\)

for 0 < y < 1. (That range is because, when x = 0, y = 0; and when x = 1, y = 1). Now, taking the derivative of v(y), we get:

\(v'(y)=\dfrac{1}{2} y^{-1/2}\)

Therefore, the change-of-variable technique:

\(f_Y(y)=f_X(v(y))\times |v'(y)|\)

tells us that the probability density function of Y is:

\(f_Y(y)=3[y^{1/2}]^2\cdot \dfrac{1}{2} y^{-1/2}\)

And, simplifying we get that the probability density function of Y is:

\(f_Y(y)=\dfrac{3}{2} y^{1/2}\)

for 0 < y < 1. We shouldn't be surprised by this result, as it is the same result that we obtained using the distribution function technique.

Example

Let's return to our example in which X is a continuous random variable with the following probability density function:

\(f(x)=3(1-x)^2\)

for 0 < x < 1. Use the change-of-variable technique to find the probability density function of  \(Y=(1-X)^3\).

Solution. Note that the function: 

\(Y=(1-X)^3\)

defined over the interval 0 < x < 1 is an invertible function. The inverse function is:

\(x=v(y)=1-y^{1/3}\)

for 0 < y < 1. (That range is because, when x = 0, y = 1; and when x = 1, y = 0). Now, taking the derivative of v(y), we get:

\(v'(y)=-\dfrac{1}{3} y^{-2/3}\)

Therefore, the change-of-variable technique:

\(f_Y(y)=f_X(v(y))\times |v'(y)|\)

tells us that the probability density function of Y is:

\(f_Y(y)=3[1-(1-y^{1/3})]^2\cdot |-\dfrac{1}{3} y^{-2/3}|=3y^{2/3}\cdot \dfrac{1}{3} y^{-2/3} \)

And, simplifying we get that the probability density function of Y is:

\(f_Y(y)=1\)

for 0 < y < 1. Again, we shouldn't be surprised by this result, as it is the same result that we obtained using the distribution function technique.