The notion of random variable and distribution of random variable

The notion of random variable is one of the most important notions of probability theory .

The variable is called random variable, if it takes values depending on results of the stochastic
experiment. This definition is not mathematically stronger.

Examples.

Example 1. Let one coin be tossed and  be the number of appearing tail. It is clear that  may
take values 0 , if head appears and 1, if tail appears. We see that this variable takes
values depending on results of experiment.

Example 2 . If coin be tossed 3 times In this case,  may take values 0, 1,2,3.

Generally, the number of success in Bernoulli scheme depends on results of experiment.

Consequently,  is random variable.

Example3 . The lifetime of one person, working time of any electronic device are random
variables.

Now we state stronger mathematical definition of random variable .

Definition. Let (Ω, ℱ, 𝑃) be any probability space . The ℱ- measurable function

=(𝜔) with real values called random variable . That is,

: Ω→ 𝑅.

The ℱ- measurability means that , for any x ∈ R

{𝜔: (𝜔) ≤ 𝑥} ∈ ℱ .

ℱ- measurability is important condition in the definition of random variable.

Next example show that.

Example. Let Ω= {𝜔1 , 𝜔2 , 𝜔3 , 𝜔4 } and ℱ be 𝜎- algebra of all subsets of Ω. The probability of

any event A∈ℱ defined as follows :
|𝐴|
P(𝐴) = .
4

Let construct the function  in Ω by the following way :

(𝜔𝑘 ) =k , k= 1,4
̅̅̅̅ .

Does this function define random variable in Ω ?

At first we see that (𝜔𝑘 ) defined on the sample space Ω . We will show that this function
ℱ- measurable . We see that values of this function devides real line into 5 intervals:

1. x< 1. In this case,{𝜔: (𝜔) ≤ 𝑥} = ∅ ∈ℱ.

2. 1≤ 𝑥 < 2 . In this case, {𝜔: (𝜔) ≤ 𝑥} = {𝜔1 } ∈ ℱ.

3. 2≤ 𝑥 < 3 . In this case, {𝜔: (𝜔) ≤ 𝑥} = {𝜔1 , 𝜔2 } ∈ ℱ.

4. 3≤ 𝑥 < 4 . In this case, {𝜔: (𝜔) ≤ 𝑥} = {𝜔1 , 𝜔2 , 𝜔3 } ∈ ℱ.

5. x≥ 4. In this case, {𝜔: (𝜔) ≤ 𝑥} = Ω ∈ ℱ.

Consequently, function  is ℱ- measurable and is random variable on sample space Ω.

Distribution Function of the Random Variable

Definition. Function 𝐹 (𝑥) = P{𝜔: (𝜔) ≤ 𝑥}, for any 𝑥 ∈ R is called distribution function of
random variable  .

Distribution function plays an important role for random variable.

We see that domain of definition of distribution function is real line R= (−∞, +∞), and the
range is [0; 1].

Properties of Distribution Function

1. Distribution function is nondecreasing , i.e., for any 𝑥1 < 𝑥2 :

𝐹(𝑥1 ) ≤ 𝐹(𝑥2 ).

Really, Let 𝑥1 < 𝑥2 . Then

{𝜉 ≤ 𝑥2 } = {𝜉 ≤ 𝑥1 } ∪ {𝑥1 < 𝜉 ≤ 𝑥2 }.

On the other hands, by additive property of probability

𝑃(𝜉 ≤ 𝑥2 ) = 𝑃(𝜉 ≤ 𝑥1 ) + 𝑃(𝑥1 < 𝜉 ≤ 𝑥2 )

Conseqvently,
 𝑃(𝜉 ≤ 𝑥2 ) − 𝑃(𝜉 ≤ 𝑥1 ) = 𝑃(𝑥1 < 𝜉 ≤ 𝑥2 ) ≥ 0

Therefore,

𝑃(𝜉 ≤ 𝑥2 ) ≥ 𝑃(𝜉 ≤ 𝑥1 ).

Taking into account

𝐹(𝑥2 ) = 𝑃(𝜉 ≤ 𝑥2 )

𝐹(𝑥1 ) = 𝑃(𝜉 ≤ 𝑥1 )

we obtain

𝐹(𝑥1 ) ≤ 𝐹(𝑥2 )

2. Distribution function is continuous from right, i.e.,

lim 𝐹(𝑥) ≡ 𝐹(𝑥0 + 0) = 𝐹(𝑥0 ).

𝑥→𝑥0 +0

3 lim 𝐹(𝑥) ≡ 𝐹(−∞) = 0 və lim 𝐹(𝑥) ≡ 𝐹(+∞) = 1.

𝑥→−∞ 𝑥→∞
 Discrete random variables and discrete distributions.

Let (Ω, ℱ, 𝑃) be any probability space and 𝜉 = 𝜉(𝜔) is random variable defined on this space.
Let 𝑥1 , … , 𝑥𝑛 , … be possible values of the random variable 𝜉. In other words, random variable 𝜉
may take finite or countable number values 𝑥1 , … , 𝑥𝑛 , ….

Let 𝑝𝑘 = 𝑃{𝜉 = 𝑥𝑘 } be corresponding probabilities of these values such that

1. 𝑝𝑘 ≥ 0; 𝑘 = 1,2, …
2. ∑∞
𝑘=1 𝑝𝑘 = 1.

In this case we say that family of probabilities {𝑝𝑘 }, 𝑘 = 1,2, … is called discrete distribution and
random variable is called discrete random variable.

Discrete distribution may be given by the form of table, graph and function.

𝑥𝑘 𝑥1 𝑥2 …

𝑝𝑘 𝑝1 𝑝2 …

table

Distribution function of the discrete random variable define by the formula

𝐹(𝑥) = ∑ 𝑝𝑘 (1)
𝑘:𝑥𝑘 ≤𝑥
This function has the following graph.

𝐹(𝑥)

𝑝3

𝑝2

𝑝1

𝑥1 𝑥2 𝑥3 𝑥

Example. Let distribution of the random variable is given:

𝑥𝑘 −1 0 2 4
𝑝𝑘 0,3 0,1 0 0,6

Find the distribution function and draw the graph.

Solution. We see that the values of the random variable

−1 0 2 4

1. 𝑥 < −1
𝐹(𝑥) = 𝑃{𝜉 ≤ 𝑥} = 𝑃{∅} = 0
2. −1 ≤ 𝑥 < 0
 𝐹(𝑥) = ∑ 𝑝𝑘 = 𝑝1 = 0,3
𝑘:𝑥𝑘 ≤𝑥

3. 0≤𝑥<2

𝐹(𝑥) = ∑ 𝑝𝑘 = 𝑝1 + 𝑝2 = 0,3 + 0,1 = 0,4

𝑘:𝑥𝑘 ≤𝑥

4. 2≤𝑥<4

𝐹(𝑥) = ∑ 𝑝𝑘 = 𝑝1 + 𝑝2 + 𝑝3 = 0,3 + 0,1 + 0 = 0,4

𝑘:𝑥𝑘 ≤𝑥

5. 𝑥4

𝐹(𝑥) = ∑ 𝑝𝑘 = 𝑝1 + 𝑝2 + 𝑝3 + 𝑝4 = 0,3 + 0,1 + 0 + 0,6 = 1

𝑘:𝑥𝑘 ≤𝑥

Some discrete distributions.

 Degenerative distribution.

𝑃{𝜉 = 𝑐} = 1, 𝑐 = 𝑐𝑜𝑛𝑠𝑡.

 Bernoulli distribution.

𝑝𝑘 = 𝑃{𝜉 = 𝑘} = 𝑝𝑘 𝑞1−𝑘 , 𝑘 = 0,1; 𝑝 + 𝑞 = 1

Or 𝑃{𝜉 = 0} = 𝑞 and 𝑃{𝜉 = 1} = 𝑝.

 Binomial distribution.

𝑃{𝜉 = 𝑘} = 𝐶𝑛𝑘 𝑝𝑘 𝑞 𝑛−𝑘 ; 𝑘 = ̅̅̅̅̅

0, 𝑛; 𝑝+𝑞 =1

This function may be discrete distribution since

∑ 𝑃{𝜉 = 𝑘} = 1
𝑘=0

𝑛 𝑛

∑ 𝑃{𝜉 = 𝑘} = ∑ 𝐶𝑛𝑘 𝑝𝑘 𝑞 𝑛−𝑘 = (𝑝 + 𝑞)𝑛 = 1

𝑘=0 𝑘=0
Note that the random variable with Binomial distribution describes the number of success in
Bernoulli scheme.

 Poisson distribution with parameter 𝜆 > 0.

𝜆𝑘 −𝜆
𝑝𝑘 = 𝑃{𝜉 = 𝑘} = 𝑒 𝑘 = 0,1,2 …
𝑘!

It is not difficult to see that 𝑝𝑘 nonnegative and

∑ 𝑃{𝜉 = 𝑘} = 1
𝑘=0

Really,

∞ ∞
𝜆𝑘 𝜆𝑘
∑ 𝑒 −𝜆 = 𝑒 −𝜆 ∑ = 𝑒 −𝜆 𝑒 𝜆 = 1
𝑘! 𝑘!
𝑘=0 𝑘=0

 Geometrical distribution.

𝑃{𝜉 = 𝑘} = 𝑝𝑞 𝑘−1 ; 𝑘 = 1,2, …

It is not difficult to see that 𝑝𝑘 nonnegative and

∑ 𝑃{𝜉 = 𝑘} = 1
𝑘=1

Really,

∞ ∞ ∞
𝑘−1
1 1
∑ 𝑃{𝜉 = 𝑘} = ∑ 𝑝𝑞 = 𝑝 ∑ 𝑞 𝑘−1 = 𝑝 = 𝑝 = 1.
1−𝑞 𝑝
𝑘=1 𝑘=1 𝑘=1

Note that random variable with geometrical distribution describe the number of independent
trails till first success is appeared.