Binomial distribution

Uses your system voice; voice and rate in settings.

Snapshot

≈ 21 min read

Names in other languages

No names in other languages are held for this entry yet.

Article bodies are English and are not machine-translated. A name in another language is not a translation of the article.

{{Probability distribution | name = Binomial distribution | type = mass | pdf_image = | cdf_image = | notation = \operatorname{B}(n,p) | parameters = n \in \{0, 1, 2, \ldots\} – number of trials p \in [0,1] – success probability for each trial q = 1 - p | support = k \in \{0, 1, \ldots, n\} – number of successes | pdf = \binom{n}{k} p^k q^{n-k} | cdf = I_q(n - \lfloor k \rfloor, 1 + \lfloor k \rfloor) (the regularized incomplete beta function) | mean = np | median = \lfloor np \rfloor or \lceil np \rceil | mode = \lfloor (n + 1)p \rfloor or \lceil (n + 1)p \rceil - 1 | variance = npq = np(1-p) | skewness = \frac{q-p}{\sqrt{npq}} | kurtosis = \frac{1-6pq}{npq} | entropy = \frac{1}{2} \log_2 (2\pi enpq) + O \left( \frac{1}{n} \right) in shannons. For nats, use the natural log in the log. | mgf = (q + pe^t)^n | char = (q + pe^{it})^n | pgf = G(z) = [q + pz]^n | fisher = g_n(p) = \frac{n}{pq} (for fixed n ) }} The probability that a ball in a Galton box with 8 layers ( ) ends up in the central bin ( ) is .

In probability theory and statistics, the binomial distribution with parameters and is the discrete probability distribution of the number of successes in a sequence of independent experiments, each asking a yes–no question, and each with its own Boolean-valued outcome: success (with probability ) or failure (with probability ).…

{{Probability distribution | name = Binomial distribution | type = mass | pdf_image = | cdf_image = | notation = \operatorname{B}(n,p) | parameters = n \in \{0, 1, 2, \ldots\} – number of trials p \in [0,1] – success probability for each trial q = 1 - p | support = k \in \{0, 1, \ldots, n\} – number of successes | pdf = \binom{n}{k} p^k q^{n-k} | cdf = I_q(n - \lfloor k \rfloor, 1 + \lfloor k \rfloor) (the regularized incomplete beta function) | mean = np | median = \lfloor np \rfloor or \lceil np \rceil | mode = \lfloor (n + 1)p \rfloor or \lceil (n + 1)p \rceil - 1 | variance = npq = np(1-p) | skewness = \frac{q-p}{\sqrt{npq}} | kurtosis = \frac{1-6pq}{npq} | entropy = \frac{1}{2} \log_2 (2\pi enpq) + O \left( \frac{1}{n} \right) in shannons. For nats, use the natural log in the log. | mgf = (q + pe^t)^n | char = (q + pe^{it})^n | pgf = G(z) = [q + pz]^n | fisher = g_n(p) = \frac{n}{pq} (for fixed n ) }} The probability that a ball in a Galton box with 8 layers ( ) ends up in the central bin ( ) is .

In probability theory and statistics, the binomial distribution with parameters and is the discrete probability distribution of the number of successes in a sequence of independent experiments, each asking a yes–no question, and each with its own Boolean-valued outcome: success (with probability ) or failure (with probability ). A single success/failure experiment is also called a Bernoulli trial or Bernoulli experiment, and a sequence of outcomes is called a Bernoulli process.

For a single trial, that is, when , the binomial distribution is a Bernoulli distribution. The binomial distribution is the basis for the binomial test of statistical significance. The binomial distribution is frequently used to model the number of successes in a sample of size drawn with replacement from a population of size .

If the sampling is carried out without replacement, the draws are not independent and so the resulting distribution is a hypergeometric distribution, not a binomial one. However, for much larger than , the binomial distribution remains a good approximation, and is widely used. f(k,n,p) = \Pr(X = k) = \binom{n}{k}p^k(1-p)^{n-k} for , where \binom{n}{k} =\frac{n!}{k!(n-k)!} is the binomial coefficient.

The formula can be understood as follows: is the probability of obtaining the sequence of independent Bernoulli trials in which trials are "successes" and the remaining trials are "failures". Since the trials are independent with probabilities remaining constant between them, any sequence of trials with successes (and failures) has the same probability of being achieved (regardless of positions of successes within the sequence).

There are \binom{n}{k} such sequences, since the binomial coefficient \binom{n}{k} counts the number of ways to choose the positions of the successes among the trials. The binomial distribution is concerned with the probability of obtaining of these sequences, meaning the probability of obtaining one of them ( ) must be added \binom{n}{k} times, hence .

In creating reference tables for binomial distribution probability, usually, the table is filled in up to values. This is because for , the probability can be calculated by its complement as f(k,n,p)=f(n-k,n,1-p). Looking at the expression as a function of , there is a value that maximizes it.

This value can be found by calculating \frac{f(k+1,n,p)}{f(k,n,p)}=\frac{(n-k)p}{(k+1)(1-p)} and comparing it to 1. There is always an integer that satisfies (n+1)p-1 \leq M is monotone increasing for and monotone decreasing for , with the exception of the case where is an integer.

In this case, there are two values for which is maximal: and . is the outcome (that is, the most likely, although this can still be unlikely overall) of the Bernoulli trials and is called the mode. f(4,6,0.3) = \binom{6}{4} 0.3^4 (1-0.3)^{6-4}= 0.059535. F(k;n,p) = \Pr(X \le k) = \sum_{i=0}^{\lfloor k \rfloor} {n\choose i}p^i(1-p)^{n-i}, where \lfloor k\rfloor is the "floor" under ; that is, the greatest integer less than or equal to .

It can also be represented in terms of the regularized incomplete beta function, as follows: \begin{align} F(k;n,p) & = \Pr(X \le k) \\ &= I_{1-p}(n-k, k+1) \\ & = (n-k) {n \choose k} \int_0^{1-p} t^{n-k-1} (1-t)^k \, dt , \end{align} which is equivalent to the cumulative distribution functions of the beta distribution and of the -distribution: F(k;n,p) = F_{\text{beta-distribution}}\left(x=1-p;\alpha=n-k,\beta=k+1\right) F(k;n,p) = F_{F\text{-distribution}}\left(x=\frac{1-p}{p}\frac{k+1}{n-k};d_1=2(n-k),d_2=2(k+1)\right). Some closed-form bounds for the cumulative distribution function are given below. \operatorname{E}[X] = np.

This follows from the linearity of the expected value along with the fact that is the sum of identical Bernoulli random variables, each with expected value . In other words, if X_1, \ldots, X_n are identical (and independent) Bernoulli random variables with parameter , then and \operatorname{E}[X] = \operatorname{E}[X_1 + \cdots + X_n] = \operatorname{E}[X_1] + \cdots + \operatorname{E}[X_n] = p + \cdots + p = np.

The variance is: \operatorname{Var}(X) = npq = np(1 - p). This similarly follows from the fact that the variance of a sum of independent random variables is the sum of the variances. The first 6 central moments, defined as \mu _{c}=\operatorname {E} \left[(X-\operatorname {E} [X])^{c}\right] , are given by \begin{align} \mu_1 &= 0, \\ \mu_2 &= np \left(1-p\right), \\ \mu_3 &= np \left(1-p\right) \left(1-2p\right), \\ \mu_4 &= np \left(1-p\right) \left[1 + \left(3n-6\right) p \left(1-p\right)\right],\\ \mu_5 &= np \left(1-p\right) \left(1-2p\right) \left[1 + \left(10n-12\right) p \left(1-p\right)\right],\\ \mu_6 &= np \left(1-p\right) \left[1 - 30p\left(1-p\right)\left[1-4p(1-p)\right] + 5np \left(1-p\right)\left[5 - 26p\left(1-p\right)\right] + 15n^2 p^2 \left(1-p\right)^2\right]. \end{align} The non-central moments satisfy \begin{align} \operatorname {E}[X] &= np, \\ \operatorname {E}[X^2] &= np(1-p)+n^2p^2, \end{align} and in general \operatorname {E}[X^c] = \sum_{k=0}^c \left\{ {c \atop k} \right\} n^{\underline{k}} p^k, where \left\{{c\atop k}\right\} are the Stirling numbers of the second kind, and n^{\underline{k}} = n(n-1)\cdots(n-k+1) is the k -th falling power of n . A simple bound follows by bounding the Binomial moments via the higher Poisson moments: \operatorname {E}[X^c] \le \left[\frac{c}{\ln\left(1+\frac{c}{np}\right)}\right]^c \le (np)^c \exp\left(\frac{c^2}{2np}\right).

This shows that if c=O(\sqrt{np}) , then \operatorname {E}[X^c] is at most a constant factor away from \operatorname {E}[X]^c . The moment-generating function is M_X(t)=\mathbb E[e^{tX}] = (1-p+p e^t)^n . \text{mode} = \begin{cases} \lfloor (n+1)\,p\rfloor & \text{if }(n+1)p\text{ is 0 or a noninteger}, \\ (n+1)\,p\ \text{ and }\ (n+1)\,p - 1 &\text{if }(n+1)p\in\{1,\dots,n\}, \\ n & \text{if }(n+1)p = n + 1. \end{cases} Proof: Let f(k)=\binom nk p^k q^{n-k}. For p=0 only f(0) has a nonzero value with f(0)=1 .

For p=1 we find f(n)=1 and f(k)=0 for k\neq n . This proves that the mode is 0 for p=0 and n for p=1 . Let 0 . We find \frac{f(k+1)}{f(k)} = \frac{(n-k)p}{(k+1)(1-p)}.

From this follows \begin{align} k > (n+1)p-1 \Rightarrow f(k+1) f(k) \end{align} So when (n+1)p-1 is an integer, then (n+1)p-1 and (n+1)p is a mode. In the case that (n+1)p-1\notin \Z , then only \lfloor (n+1)p-1\rfloor+1=\lfloor (n+1)p\rfloor is a mode.

When p = \tfrac{1}{2} and is odd, any number in the interval \frac{1}{2} \left(n-1\right) \leq m \leq \frac{1}{2} \left(n+1\right) is a median of the binomial distribution. If p = \tfrac{1}{2} and is even, then m = \tfrac{n}{2} is the unique median.

For , upper bounds can be derived for the lower tail of the cumulative distribution function F(k;n,p) = \Pr(X \le k) , the probability that there are at most successes. Since \Pr(X \ge k) = F(n-k;n,1-p) , these bounds can also be seen as bounds for the upper tail of the cumulative distribution function for .

Hoeffding's inequality yields the simple bound F(k;n,p) \leq \exp\left(-2 n\left(p-\frac{k}{n}\right)^2\right), \! which is however not very tight. In particular, for , we have that (for fixed , with ), but Hoeffding's bound evaluates to a positive constant.

A sharper bound can be obtained from the Chernoff bound: F(k;n,p) \leq \exp\left(-n D{\left(\frac{k}{n}\parallel p\right)}\right) where is the relative entropy (or Kullback-Leibler divergence) between an -coin and a -coin (that is, between the and distribution): D(a\parallel p)=(a)\ln\frac{a}{p}+(1-a)\ln\frac{1-a}{1-p}. \! Asymptotically, this bound is reasonably tight; see for details.

One can also obtain lower bounds on the tail , known as anti-concentration bounds. By approximating the binomial coefficient with Stirling's formula it can be shown that F(k;n,p) \geq \frac{1}{\sqrt{8n\tfrac{k}{n}(1-\tfrac{k}{n})}} \exp\left(-n D{\left(\frac{k}{n}\parallel p\right)}\right), which implies the simpler but looser bound F(k;n,p) \geq \frac1{\sqrt{2n}} \exp\left(-nD\left(\frac{k}{n}\parallel p\right)\right).

For and for even , it is possible to make the denominator constant: F(k;n,\tfrac{1}{2}) \geq \frac{1}{15} \exp\left(- 16n \left(\frac{1}{2} -\frac{k}{n}\right)^2\right). \! \widehat{p} = \frac{x}{n}. This estimator is found using maximum likelihood estimator and also the method of moments.

This estimator is unbiased and uniformly with minimum variance, proven using Lehmann–Scheffé theorem, since it is based on a minimal sufficient and complete statistic (that is, ). It is also consistent both in probability and in MSE. This statistic is asymptotically normal thanks to the central limit theorem, because it is the same as taking the mean over Bernoulli samples.

It has a variance of \operatorname{Var}(\hat{p}) = \frac{p(1-p)}{n} , a property which is used in various ways, such as in Wald's confidence intervals. A closed form Bayes estimator for also exists when using the Beta distribution as a conjugate prior distribution.

When using a general \operatorname{Beta}(\alpha, \beta) as a prior, the posterior mean estimator is: \widehat{p}_b = \frac{x+\alpha}{n+\alpha+\beta}. The Bayes estimator is asymptotically efficient and as the sample size approaches infinity ( ), it approaches the MLE solution. The Bayes estimator is biased (how much depends on the priors), admissible and consistent in probability.

Using the Bayesian estimator with the Beta distribution can be used with Thompson sampling. For the special case of using the standard uniform distribution as a non-informative prior, \operatorname{Beta}(\alpha{=}1,\, \beta{=}1) = U(0,1) , the posterior mean estimator becomes: \widehat{p}_b = \frac{x+1}{n+2}. (A posterior mode should just lead to the standard estimator.)

This method is called the rule of succession, which was introduced in the 18th century by Pierre-Simon Laplace. When relying on Jeffreys prior, the prior is \operatorname{Beta}(\alpha{=}\tfrac{1}{2}, \, \beta{=}\tfrac{1}{2}) , which leads to the estimator: \widehat{p}_{\mathrm{Jeffreys}} = \frac{x+\frac{1}{2}}{n+1}.

When estimating with very rare events and a small (for example, if ), then using the standard estimator leads to \widehat{p} = 0, which sometimes is unrealistic and undesirable. In such cases there are various alternative estimators. One way is to use the Bayes estimator \widehat{p}_b , leading to: \widehat{p}_b = \frac{1}{n+2}.

Another method is to use the upper bound of the confidence interval obtained using the rule of three: \widehat{p}_{\text{rule of 3}} = \frac{3}{n}. \widehat{p\,} = \frac{n_1}{n} is the proportion of successes z is the 1 - \tfrac{1}{2}\alpha quantile of a standard normal distribution (that is, probit) corresponding to the target error rate \alpha . For example, for a 95% confidence level the error \alpha=0.05 , so 1 - \tfrac{1}{2}\alpha=0.975 and z=1.96 . \widehat{p\,} \pm z \sqrt{ \frac{ \widehat{p\,} ( 1 -\widehat{p\,} )}{ n } } .

A continuity correction of may be added. \tilde{p} \pm z \sqrt{ \frac{ \tilde{p} ( 1 - \tilde{p} )}{ n + z^2 } } Here the estimate of is modified to \tilde{p}= \frac{ n_1 + \frac{1}{2} z^2}{ n + z^2 } This method works well for and . See here for n\leq 10 . For use the Wilson (score) method below. \sin^2 \left(\arcsin \left(\sqrt{\hat{p}}\right) \pm \frac{z}{2\sqrt{n}} \right).

Secondly, this formula does not use a plus-minus to define the two bounds. Instead, one may use z = z_{\alpha / 2} to get the lower bound, or use z = z_{1 - \alpha/2} to get the upper bound.

For example: for a 95% confidence level the error \alpha=0.05 , so one gets the lower bound by using z = z_{\alpha/2} = z_{0.025} = - 1.96 , and one gets the upper bound by using z = z_{1 - \alpha/2} = z_{0.975} = 1.96 . \frac{ \hat{p} + \frac{z^2}{2n} + z \sqrt{ \frac{\hat{p} \left(1 - \hat{p}\right)}{n} + \frac{z^2}{4 n^2} } }{ 1 + \frac{z^2}{n} } \begin{align} \operatorname P(Z=k) &= \sum_{i=0}^k\left[\binom{n}i p^i (1-p)^{n-i}\right]\left[\binom{m}{k-i} p^{k-i} (1-p)^{m-k+i}\right]\\ &= \binom{n+m}k p^k (1-p)^{n+m-k} \end{align} A Binomial distributed random variable can be considered as the sum of Bernoulli distributed random variables. So the sum of two Binomial distributed random variables and is equivalent to the sum of Bernoulli distributed random variables, which means .

This can also be proven directly using the addition rule. However, if and do not have the same probability , then the variance of the sum will be smaller than the variance of a binomial variable distributed as . \begin{align} \Pr[Y = m] &= \sum_{k = m}^{n} \Pr[Y = m \mid X = k] \Pr[X = k] \\[2pt] &= \sum_{k=m}^n \binom{n}{k} \binom{k}{m} p^k q^m (1-p)^{n-k} (1-q)^{k-m} \end{align} Since \tbinom{n}{k} \tbinom{k}{m} = \tbinom{n}{m} \tbinom{n-m}{k-m}, the equation above can be expressed as \Pr[Y = m] = \sum_{k=m}^{n} \binom{n}{m} \binom{n-m}{k-m} p^k q^m (1-p)^{n-k} (1-q)^{k-m} Factoring p^k = p^m p^{k-m} and pulling all the terms that don't depend on k out of the sum now yields \begin{align} \Pr[Y = m] &= \binom{n}{m} p^m q^m \left( \sum_{k=m}^n \binom{n-m}{k-m} p^{k-m} (1-p)^{n-k} (1-q)^{k-m} \right) \\[2pt] &= \binom{n}{m} (pq)^m \left( \sum_{k=m}^n \binom{n-m}{k-m} \left(p(1-q)\right)^{k-m} (1-p)^{n-k} \right) \end{align} After substituting i = k - m in the expression above, we get \Pr[Y = m] = \binom{n}{m} (pq)^m \left( \sum_{i=0}^{n-m} \binom{n-m}{i} (p - pq)^i (1-p)^{n-m - i} \right) Notice that the sum (in the parentheses) above equals (p - pq + 1 - p)^{n-m} by the binomial theorem.

Substituting this in finally yields \begin{align} \Pr[Y=m] &= \binom{n}{m} (pq)^m (p - pq + 1 - p)^{n-m}\\[4pt] &= \binom{n}{m} (pq)^m (1-pq)^{n-m} \end{align} and thus Y \sim \mathrm{B}(n, pq) as desired. \mathcal{N}(np,\,np(1-p)), and this basic approximation can be improved in a simple way by using a suitable continuity correction. The basic approximation generally improves as increases (at least 20) and is better when is not near to 0 or 1.

Various rules of thumb may be used to decide whether is large enough, and is far enough from the extremes of zero or one: One rule is that for the normal approximation is adequate if the absolute value of the skewness is strictly less than 0.3; that is, if \frac {\sqrt{np(1-p)}}=\frac1{\sqrt{n}}\left|\sqrt{\frac{1-p}p}-\sqrt{\frac{p}{1-p}}\,\right| This can be made precise using the Berry–Esseen theorem. A stronger rule states that the normal approximation is appropriate only if everything within 3 standard deviations of its mean is within the range of possible values; that is, only if \mu \pm 3\sigma = n p \pm 3 \sqrt{np(1-p)}\in(0,n). This 3-standard-deviation rule is equivalent to the following conditions, which also imply the first rule above. n>9 \left(\frac{1-p}{p} \right)\quad\text{and}\quad n>9\left(\frac{p}{1-p}\right).

The rule np\pm3\sqrt{np(1-p)}\in(0,n) is totally equivalent to request that np-3\sqrt{np(1-p)}>0\quad\text{and}\quad np+3\sqrt{np(1-p)} Moving terms around yields: np>3\sqrt{np(1-p)}\quad\text{and}\quad n(1-p)>3\sqrt{np(1-p)}. Since 0 , we can apply the square power and divide by the respective factors np^2 and n(1-p)^2 , to obtain the desired conditions: n>9 \left(\frac{1-p}p\right) \quad\text{and}\quad n>9 \left(\frac{p}{1-p}\right).

Notice that these conditions automatically imply that n>9 . On the other hand, apply again the square root and divide by 3, \frac{\sqrt{n}}3>\sqrt{\frac{1-p}p}>0 \quad \text{and} \quad \frac{\sqrt{n}}3 > \sqrt{\frac{p}{1-p}}>0.

Subtracting the second set of inequalities from the first one yields: \frac{\sqrt{n}}3>\sqrt{\frac{1-p}p}-\sqrt{\frac{p}{1-p}}>-\frac{\sqrt{n}}3; and so, the desired first rule is satisfied, \left|\sqrt{\frac{1-p}p}-\sqrt{\frac{p}{1-p}}\,\right| Another commonly used rule is that both values and must be greater than or equal to 5. However, the specific number varies from source to source, and depends on how good an approximation one wants.

In particular, if one uses 9 instead of 5, the rule implies the results stated in the previous paragraphs. Assume that both values np and n(1-p) are greater than 9.

Since 0 , we easily have that np\geq9>9(1-p)\quad\text{and}\quad n(1-p)\geq9>9p. We only have to divide now by the respective factors p and 1-p , to deduce the alternative form of the 3-standard-deviation rule: n>9 \left(\frac{1-p}p\right) \quad\text{and}\quad n>9 \left(\frac{p}{1-p}\right). The following is an example of applying a continuity correction.

Suppose one wishes to calculate for a binomial random variable . If has a distribution given by the normal approximation, then is approximated by . The addition of 0.5 is the continuity correction; the uncorrected normal approximation gives considerably less accurate results.

This approximation, known as de Moivre–Laplace theorem, is a huge time-saver when undertaking calculations by hand (exact calculations with large are very onerous); historically, it was the first use of the normal distribution, introduced in Abraham de Moivre's book The Doctrine of Chances in 1738. Nowadays, it can be seen as a consequence of the central limit theorem since is a sum of independent, identically distributed Bernoulli variables with parameter  .

This fact is the basis of a hypothesis test, a "proportion z-test", for the value of using , the sample proportion and estimator of , in a common test statistic. For example, suppose one randomly samples people out of a large population and ask them whether they agree with a certain statement. The proportion of people who agree will of course depend on the sample.

If groups of people were sampled repeatedly and truly randomly, the proportions would follow an approximate normal distribution with mean equal to the true proportion of agreement in the population and with standard deviation \sigma = \sqrt{\frac{p(1-p)}{n}} de Moivre–Laplace theorem: As approaches while remains fixed, the distribution of \frac{X-np}{\sqrt{np(1-p)}} approaches the normal distribution with expected value 0 and variance 1. This result is sometimes loosely stated by saying that the distribution of is asymptotically normal with expected value 0 and variance 1.

This result is a specific case of the central limit theorem. \operatorname{Beta}(p;\alpha;\beta) = (n+1)\mathrm{B}(k;n;p) Beta distributions also provide a family of prior probability distributions for binomial distributions in Bayesian inference: P(p;\alpha,\beta) = \frac{p^{\alpha-1}(1-p)^{\beta-1}}{\operatorname{Beta}(\alpha,\beta)}. Given a uniform prior, the posterior distribution for the probability of success given independent events with observed successes is a beta distribution.

Querying the binomial probability distribution in WolframAlpha Confidence (credible) intervals for binomial probability, p: online calculator available at causaScientia.org

Text derived from https://en.wikipedia.org/w/index.php?oldid=1374408439 under CC BY-SA 4.0; modifications have been made.

Sources cited in this revision

27 named source(s) extracted from the cited revision. A source with no author is shown by its publication; nothing is invented.