24.03.2013 Views

Linear Transformations and Combinations

Linear Transformations and Combinations

Linear Transformations and Combinations

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

person buys, X, is a r<strong>and</strong>om variable. Suppose we want an expression for the amount of money<br />

spent by that particular person. Letting Y denote the amount of money spent, we can find a<br />

function relating Y to X. If a person buys X tickets, then they spent 2X dollars on tickets. In<br />

addition, they spent $5 to enter the carnival. The total amount spent is Y = 2X + 5. Since Y has<br />

the correct form (aX + b, with a = 2, <strong>and</strong> b = 5), we say Y is a linear transformation of X.<br />

If we know the mean <strong>and</strong> variance of X, then there are simple formulas to compute the mean<br />

<strong>and</strong> variance of a linear transformation Y . We derive the formulas for discrete r<strong>and</strong>om variables.<br />

We know that E[X] = <br />

<br />

x xP (X = x) <strong>and</strong> E[h(X)] = x h(x)P (X = x). A linear transformation<br />

is a particular form of h(x), so<br />

The formula applies because E[X] = <br />

x<br />

r<strong>and</strong>om variable.<br />

E[aX + b] = <br />

x (ax + b)P (X = x)<br />

= <br />

x axP (X = x) + bP (X = x)<br />

= <br />

<br />

x axP (X = x) + x bP (X = x)<br />

= a <br />

<br />

x xP (X = x) + b x P (X = x)<br />

= aE[X] + b<br />

<br />

xP (X = x) by definition <strong>and</strong> x P (X = x) = 1 for any<br />

V [aX + b] = E[((aX + b) − (aE[X] + b)) 2 ]<br />

= E[(aX + b − aE[X] − b) 2 ]<br />

= E[a 2 (X − E[X]) 2 ]<br />

= a 2 E[(X − E[X]) 2 ]<br />

= a 2 V [X]<br />

These formulas E[aX + b] = aE[X] + b <strong>and</strong> V [aX + b] = a 2 V [X] apply for any r<strong>and</strong>om variable,<br />

discrete or continuous (we did not derive the calculus, but integrals have many of the properties of<br />

summations, such as factoring out constants <strong>and</strong> breaking sums into two parts, so it makes sense).<br />

The expectation formula is simple to remember, the expectation of a linear transformation is<br />

the same linear transformation of the expectation. The variance requires a little more explanation.<br />

Variance is a measure of spread. Adding a constant b doesn’t change the spread, so it can be<br />

ignored in computed the variance. When we multiply by a, we have to remember the variance is<br />

in squared units, so the constant a is squared in the formula.<br />

Suppose for our carnival example any particular person buys an average of 5 tickets with a<br />

variance of 9 tickets. What is the mean <strong>and</strong> variance of the amount of money spent by a particular<br />

person? We said Y = 2X + 5, so E[Y ] = 2E[X] + 5 = 2(5) + 5 = 15 <strong>and</strong> V [Y ] = 2 2 (9) = 36.<br />

3 <strong>Linear</strong> <strong>Combinations</strong><br />

Often we deal with several r<strong>and</strong>om variables at once. A linear combination of two r<strong>and</strong>om variables<br />

X <strong>and</strong> Y has the form aX + bY + c, where a, b, <strong>and</strong> c are fixed constants found from the problem.<br />

<strong>Linear</strong> combinations can involve many r<strong>and</strong>om variables. A linear combination Y of n r<strong>and</strong>om<br />

variables X1, . . . , Xn is<br />

Y = a1X1 + a2X2 + . . . + anXn + b<br />

where the ai <strong>and</strong> b coefficients are fixed values which, again, are found in the problem.<br />

2

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!