A random variable X has a discrete uniform distribution if it can take on a finite number of equally likely values.
Let X take values {a, a+1, a+2, ..., b}
, where a
and b
are integers with a β€ b
. Then:
n = b β a + 1
f(x) = 1 / n
for x β {a, ..., b}
By definition: E(X) = Ξ£ xΒ·f(x)
Since each f(x) = 1/n
, we have:
E(X) = (1/n) Β· Ξ£ from x=a to b of x = (1/n) Β· [ (a + b)(n) / 2 ] = (a + b) / 2
βοΈ The mean is simply the midpoint of a
and b
.
First, compute E(XΒ²)
:
E(XΒ²) = (1/n) Β· Ξ£ from x=a to b of xΒ²
Using the formula for the sum of squares:
Ξ£_{k=1}^m kΒ² = m(m+1)(2m+1)/6
,
we can express E(XΒ²)
in closed form.
After simplification, the variance of a discrete uniform distribution is:
Var(X) = [(nΒ² β 1)] / 12 where n = (b β a + 1)
Ο = βVar(X) = β( (nΒ² β 1) / 12 )
ΞΌ = (a + b) / 2
ΟΒ² = ( (b β a + 1)Β² β 1 ) / 12
Ο = β[ ( (b β a + 1)Β² β 1 ) / 12 ]
The first digit of a part's serial number is equally likely to be any digit from 0 to 9.
This gives a uniform distribution with 10 values, each with probability 0.1.
Let X be the number of voice lines in use at a given time. X is uniformly distributed over {0, 1, ..., 48}.
Interpretation: On average, 24 voice lines are in use. However, the high standard deviation means that usage can vary widely, sometimes far above or below the average.
Let X be the number of voice lines used out of 48, and define a new variable Y = X / 48, the proportion of lines used. Since X is uniformly distributed over {0, 1, ..., 48}, we can use linear transformation formulas:
Explanation: Expectation is linear, so scaling X by 1/48 scales the mean by 1/48. Variance scales by the square of the factor: Var(aX) = aΒ²Var(X).
β Even though the average proportion is 50%, the standard deviation tells us that actual usage might often fall between roughly 20.5% and 79.5%.
π This kind of transformation is important when comparing different systems with different total capacities using normalized units like percentages or proportions.
X ~ Uniform(1, 7). Find:
ΞΌ = (1 + 7) / 2 = 4
ΟΒ² = [(7 β 1 + 1)Β² β 1] / 12 = (7Β² β 1)/12 = 48 / 12 = 4
Ο = β4 = 2