Log-normal distribution
|
|
Template:Probability distribution
In probability and statistics, the log-normal distribution is the probability distribution of any random variable whose logarithm is normally distributed (the base of the logarithmic function is immaterial in that loga X is normally distributed if and only if logb X is normally distributed). If X is a random variable with a normal distribution, then exp(X) has a log-normal distribution.
"Log-normal" is also written "log normal" or "lognormal".
A variable might be modeled as log-normal if it can be thought of as the multiplicative product of many small independent factors. A typical example is the long-term return rate on a stock investment: it can be considered as the product of the daily return rates.
The log-normal distribution has probability density function
- <math>f(x|\mu,\sigma) = \frac{1}{x \sigma \sqrt{2 \pi}} e^{-(\ln x - \mu)^2/2\sigma^2}<math>
for <math>x>0<math>, where <math>\mu<math> and <math>\sigma<math> are the mean and standard deviation of the variable's logarithm. The expected value is
- <math>\mathrm{E}(X) = e^{\mu + \sigma^2/2}<math>
and the variance is
- <math>\mathrm{var}(X) = (e^{\sigma^2} - 1) e^{2\mu + \sigma^2}<math>.
| Contents |
Relationship to geometric mean and geometric standard deviation
The log-normal distribution, the geometric mean, and the geometric standard deviation are related. In this case, the geometric mean is equal to <math>\exp(\mu)<math> and the geometric standard deviation is equal to <math>\exp(\sigma)<math>.
If a sample of data is determined to come from a log-normally distributed population, the geometric mean and the geometric standard deviation may be used to estimate confidence intervals akin to the way the arithmetic mean and standard deviation are used to estimate confidence intervals for a normally distributed sample of data.
| Confidence interval bounds | log space | geometric |
|---|---|---|
| 3σ lower bound | <math>\mu - 3\sigma<math> | <math>\mu_{geo} / \sigma_{geo}^3<math> |
| 2σ lower bound | <math>\mu - 2\sigma<math> | <math>\mu_{geo} / \sigma_{geo}^2<math> |
| 1σ lower bound | <math>\mu - \sigma<math> | <math>\mu_{geo} / \sigma_{geo}<math> |
| 1σ upper bound | <math>\mu + \sigma<math> | <math>\mu_{geo} \sigma_{geo}<math> |
| 2σ upper bound | <math>\mu + 2\sigma<math> | <math>\mu_{geo} \sigma_{geo}^2<math> |
| 3σ upper bound | <math>\mu + 3\sigma<math> | <math>\mu_{geo} \sigma_{geo}^3<math> |
Where geometric mean <math>\mu_{geo} = \exp(\mu)<math> and geometric standard deviation <math>\sigma_{geo} = \exp(\sigma)<math>
Moments
The first few raw moments are:
- <math>\mu_1=e^{\mu+\sigma^2/2}<math>
- <math>\mu_2=e^{2\mu+4\sigma^2/2}<math>
- <math>\mu_3=e^{3\mu+9\sigma^2/2}<math>
- <math>\mu_4=e^{4\mu+16\sigma^2/2}<math>
or generally:
- <math>\mu_k=e^{k\mu+k^2\sigma^2/2}<math>
Maximum likelihood estimation of parameters
For determining the maximum likelihood estimators of the log-normal distribution parameters μ and σ, we can use the same procedure as for the normal distribution. To avoid repetition, we observe that
- <math>f_L (x|\mu, \sigma) = \frac 1 x \, f_N (\ln x| \mu, \sigma)<math>
where by <math>f_L (\cdot)<math> we denote the density probability function of the log-normal distribution and by <math>f_N (\cdot)<math>—that of the normal distribution. Therefore, using the same indices to denote distributions, we can write that
- <math>\begin{matrix}
\ell_L (x_1, x_2, ..., x_n| \mu, \sigma) & = & - \sum _k \ln x_k + \ell_N (\ln x_1, \ln x_2, ..., \ln x_n| \mu, \sigma) = \\
\ & = & \operatorname {const} (\mu, \sigma) + \ell_N (\ln x_1, \ln x_2, ..., \ln x_n| \mu, \sigma). \end{matrix}<math>
Since the first term is constant with regards to μ and σ, both logarithmic likelihood functions, <math>\ell_L<math> and <math>\ell_N<math>, reach their maximum with the same μ and σ. Hence, using the formulas for the normal distribution maximum likelihood parameter estimators and the equality above, we deduce that for the log-normal distribution it holds that
- <math>\hat \mu = \frac {\sum_k \ln x_k} n, \
\hat \sigma = \frac {\sum_k {\left( \ln x_k - \hat \mu \right)^2}} n.<math>
Related distributions
- <math>Y \sim N(\mu, \sigma^2)<math> is a normal distribution if <math>Y = \ln(X)<math> and <math>X \sim \operatorname{Log-N}(\mu, \sigma^2)<math>.
- If <math>X_m \sim \operatorname {Log-N} (\mu, \sigma_m^2), \ m = \overline {1 ... N}<math> are independent log-normally distributed variables with the same μ parameter and possibly varying σ, and <math>Y = \prod_{m=1}^N X_m<math>, then Y is a log-normally distributed variable as well: <math>Y \sim \operatorname {Log-N} \left( \mu, \sum _m \sigma_m^2 \right)<math>.