Вы находитесь на странице: 1из 1

I'm assuming you mistyped the problem, where the second expression is for y between 5 and 10 rather than

0 and 10. Else you have different expression for interval 0 to 5... anyways, a) First obtain the cdf of Y, you need to break up the function into 2 parts as well, noting integrating f(y) from 0 to 5 is 0.5. so for y between 0,5 you integrate 0 to y of t/25 => F(y) = y^2/50, for y between 0 and 5 and for y between 5 and 10, you integrate 5 to y of 2/5-t/25 and plus 0.5 (for integrating 0 to 5 of f(y)), which is: 2y/5 - y^2/50 - 3/2 + 1/2 = 2y/5 - y^2/50 - 1 so your cdf is defined as: F(y) = y^2/50 if 0<y<5 = 2y/5 - y^2/50 - 1 if 5<y<10 (strictly less than or less than/equal to is same for continuous pdf, so I'm a bit sloppy with notations here) and to obtain (100p)th percentile, you simply set F(y)=p and solve. Again your functions are defined in two parts, so you need to solve for two expressions of p. When y is between 0 and 5, p is between 0 and 0.5; and when y is between 5 and 10, p is between 0.5 and 1. b) E(Y) = integrate from 0 to 5 of y*y/25 + integrate from 5 to 10 of y*[(2/5)-(1/25)y] = 5/3+10/3 = 5 V(Y) = E(Y^2) - [E(Y)]^2, where E(Y^2) =integrate from 0 to 5 of y^2*y/25 + integrate from 5 to 10 of y^2*[(2/5)-(1/25)y] = 25/4+275/12 = 175/6 = 29.16667 so V(Y) = 29.16667 - 5^2 = 4.1667 For the uniform distribution (call the variable X), X~U(0,5) E(X) = (5-0)/2 = 2.5, and V(X) = (5-0)^2/12 = 25/12 = 2.0833 and if you compare the results, E(Y) = 2*E(X) and V(Y) = 2*V(X)

Вам также может понравиться