1.2.2. Binomial

ESTADÍSTICA

VARIABLE ALEATORIA BINOMIAL  

Se basa un experimento subyacente de Bernouilli, es decir, un suceso   que puede suceder (éxito) o no (fracaso). La variable binomial consiste en repetir el experimento de Bernouilli cierto número de veces y registrar el número de éxitos ocurridos. Cada repetición es independiente de las demás.  

Parámetro/s que define esta variable

Número de repeticiones

• n = 20

Probabilidad de éxito de la v.a. de Bernouilli

• p = 0.1

Valor variable(k)	FunciónProbabilidadpX(k)	kpX(k)	k2pX(k)
0	0.1215766546	0	0
1	0.2701703435	0.2701703435	0.2701703435
2	0.2851798071	0.5703596141	1.140719228
3	0.1901198714	0.5703596141	1.711078842
4	0.08977882815	0.3591153126	1.43646125
5	0.03192136112	0.1596068056	0.798034028
6	0.008867044756	0.05320226853	0.3192136112
7	0.00197045439	0.01379318073	0.09655226512
8	0.0003557764871	0.002846211897	0.02276969517
9	0.00005270762772	0.0004743686495	0.004269317845
10	0.000006442043388	0.00006442043388	0.0006442043388
11	0.0000006507114533	0.000007157825987	0.00007873608585
12	0.00000005422595444	0.0000006507114533	0.00000780853744
13	0.000000003707757569	0.00000004820084839	0.0000006266110291
14	0.0000000002059865316	0.000000002883811442	0.00000004037336019
15	0	0.0000000001373243544	0.000000002059865316
16	0	0	0
17	0	0	0
18	0	0	0
19	0	0	0
𝚺	1	2	5.8
𝚺²	1	4	33.64
𝚿	0.05	0.1	0.29
𝚫	0.2851798071	0.5703596141	1.711078842
𝚸²	0.2144078332	0.8810793969	7.114059188
𝚸	0.4630419346	0.9386582961	2.667219374
	Calculada	Teórica
ESPERANZA	1.999999949	2
VARIANZA	1.799999536	1.8

---

title: "Binomial Distribution"

The Binomial distribution is one of the cleanest models in statistics.

It belongs to a world of repeated trials, each with only two possible outcomes:

• success / failure
• yes / no
• hit / miss
• pass / fail
• defect / non-defect

From this simple structure, a rich theory emerges.

The Binomial is not merely a counting device.
It is the statistical form of repeated discrete uncertainty.

---

🏵 The setup

Suppose we repeat the same trial $n$ times.

Each trial has:

• two possible outcomes
• the same probability of success $p$
• independence from the others

Then the number of successes $X$ follows a Binomial distribution:

$$ X \sim \mathrm{Binomial}(n,p) $$

The probability of exactly $k$ successes is:

$$ \mathbb{P}(X = k)

\binom{n}{k} p^k (1-p)^{n-k} $$

This formula combines two ideas:

• how many ways $k$ successes can be arranged
• the probability of each such arrangement

---

🔰 Why it matters

The Binomial matters because it transforms a sequence of uncertain events into a count.

Instead of asking:

what happened on each trial?

it asks:

how many times did success occur?

This is a statistical compression.

Order is forgotten.
Count is preserved.

And often, count is exactly what matters.

Examples:

• number of heads in coin tosses
• number of defective items in a batch
• number of correct answers on a test
• number of patients responding to treatment
• number of emails opened in a campaign

---

☯ Parameters

The Binomial has two parameters:

Number of trials: $n$

This determines how many opportunities for success there are.

Probability of success: $p$

This determines how likely each success is.

Together, these shape the distribution.

If $p = 0.5$, the distribution is symmetric when $n$ is not too small.
If $p$ is small or large, the distribution becomes skewed.

So even within the Binomial family, shape depends strongly on the underlying chance of success.

---

⚙️ Mean and variance

For a Binomial variable:

$$ \mathbb{E}[X] = np $$

and

$$ \mathrm{Var}(X) = np(1-p) $$

These formulas are compact but revealing.

The expected number of successes is simply:

• number of chances
• times success probability

The variance depends on both success and failure.

That is beautiful.

There is no spread if success is impossible or certain.
Variation is greatest somewhere in the uncertain middle.

This expresses a deep truth:

uncertainty is largest when the process is neither locked into failure nor locked into success

---

💡 Independence and sameness

The Binomial rests on strong assumptions:

• trials are independent
• each trial has the same probability $p$

These assumptions make the model elegant, but they are not always realistic.

Real processes may involve:

• fatigue
• learning
• contagion
• changing conditions
• feedback
• dependence

So the Binomial is not a universal model of repeated trials.
It is the clean ideal case.

That is precisely why it is so useful: it gives a baseline world against which more complex reality can be compared.

---

🪄 A bridge between probability and statistics

The Binomial sits beautifully between probability and statistics.

On the probability side, it is a model of repeated chance.
On the statistical side, it becomes a natural model for data involving proportions and counts.

From the Binomial grow ideas such as:

• sample proportion
• standard error of a proportion
• approximation by Normal methods
• hypothesis tests for proportions
• confidence intervals for binary outcomes

So this distribution is small in setup and large in consequences.

---

⚠️ Limits of the Binomial

The Binomial should not be used carelessly.

It fails when:

• trials are not independent
• success probability changes across trials
• outcomes are not truly binary
• counts arise from more open-ended arrival processes
• the mechanism involves clustering or contagion

In such cases, other models may be better:

• Negative Binomial
• Poisson
• Beta-Binomial
• Markov-style dependence structures

But the Binomial remains the natural first grammar of repeated yes/no uncertainty.

---

🔰 The Binomial as structure

What makes the Binomial so satisfying is its simplicity.

It takes:

• uncertainty
• repetition
• binary outcome
• count

and joins them into one object.

This is one of the recurring beauties of statistics:

a very simple formal structure can illuminate a wide range of real situations.

---

🏵 Final thought

The Binomial distribution is the arithmetic of repeated possibility.

It teaches that uncertainty need not be continuous, vague, or mysterious.

Sometimes it arrives in the clearest possible form:

• one trial
• then another
• then another
• and finally, a count

And from that count, a great deal of statistical reasoning begins.