Unit Gompertz distribution

{{Probability distribution | name = unit-Gompertz Distribution | type = density | pdf_image = | cdf_image = | parameters = \alpha > 0\, (real) \beta > 0\, (real) | support = x \in (0,1)\, | pdf = \alpha \beta x^{-(\beta+1)} \exp \left[…

Uses your system voice; voice and rate in settings.

Snapshot

≈ 5 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 = unit-Gompertz Distribution | type = density | pdf_image = | cdf_image = | parameters = \alpha > 0\, (real) \beta > 0\, (real) | support = x \in (0,1)\, | pdf = \alpha \beta x^{-(\beta+1)} \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right] | cdf = \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right] | quantile = \exp \left[ -\frac{1}{\beta} \log(\alpha - \log p) - \log \alpha \right] | mode = x_0 = \left( \frac{\alpha \beta}{\beta + 1} \right)^{\frac{1}{\beta}} | mean = \alpha^{\frac{1}{\beta}} e^\alpha \Gamma \left( 1 - \frac{1}{\beta}, \alpha \right) | variance = \alpha^{\frac{2}{\beta}} e^{2\alpha} \left\{ \Gamma \left( 1 - \frac{2}{\beta}, \alpha \right) - \left[ \Gamma \left( 1 - \frac{1}{\beta}, \alpha \right) \right]^2 \right\} | mgf = \alpha^{\frac{r}{\beta}} e^\alpha \Gamma \left( 1 - \frac{r}{\beta}, \alpha \right) | skewness = \frac{\mu'_3 - 3\mu'_2\mu + \mu^3}{\sigma^3} | kurtosis = \frac{\mu'_4 - 4\mu'_3\mu + 6\mu'_2\mu^2 - 3\mu^4}{\sigma^4} | }} The unit-Gompertz distribution (UGo) is a continuous probability distribution with domain on (0,1) . Useful for bounded variables with a (0,1) domain.

It was originally proposed by Mazucheli et al using a transformation of the Gompertz distribution. It is a suitable alternative for fitting skewed data. A quantile regression model has also been derived from the distribution. f(x \mid \alpha, \beta) = \alpha \beta x^{-(\beta+1)} \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right].

F(x \mid \alpha, \beta) = \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right].

Quantile function

The quantile function of the UGo distribution is given by: Q(p \mid \alpha, \beta) = \exp \left[ -\frac{1}{\beta} \log(\alpha - \log p) - \log \alpha \right]. \mu'_r = \mathbb{E}(X^r) = \int_0^1 x^r \alpha \beta x^{-(\beta+1)} \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right] dx = \alpha^{\frac{r}{\beta}} e^\alpha \Gamma \left( 1 - \frac{r}{\beta}, \alpha \right), where \Gamma(\cdot, \cdot) is the upper incomplete gamma function. The moments exists only when \frac{r}{\beta} . \mathit{skewness} = \frac{\mu'_3 - 3\mu'_2 \mu + \mu^3}{\sigma^3}, \mathit{kurtosis} = \frac{\mu'_4 - 4\mu'_3 \mu + 6\mu'_2 \mu^2 - 3\mu^4}{\sigma^4} h(x \mid \alpha, \beta) = \frac{\alpha \beta x^{-(\beta+1)} \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right]}{1 - \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right]}, \quad 0 Let \mathbf{x} = (x_1, \ldots, x_n) be a random sample of size n from the UGo distribution with probability density function defined before.

Then, the log-likelihood function of \boldsymbol{\theta} = (\alpha, \beta) is: \begin{align} \ell(\mathbf{x} \mid \boldsymbol{\theta}) = n \log \alpha + n \log \beta - (\beta + 1) \sum_{i=1}^n \log x_i - \alpha \sum_{i=1}^n x_i^{-\beta} + n \alpha. \end{align} The likelihood estimate \hat{\boldsymbol{\theta}} of \boldsymbol{\theta} is obtained by solving the non-linear equations \frac{\partial \ell}{\partial \alpha} = n \left( 1 + \frac{1}{\alpha} \right) - \sum_{i=1}^n x_i^{-\beta}, and \frac{\partial \ell}{\partial \beta} = \frac{n}{\beta} + \alpha \sum_{i=1}^n \frac{\log x_i}{x_i^\beta} - \sum_{i=1}^n \log x_i. The Hessian matrix of the log-likelihood with respect to the unknown parameters is given by \mathbf{H} = \begin{bmatrix} -\dfrac{n}{\alpha^2} & \vdots & \displaystyle\sum_{i=1}^n \frac{\log x_i}{x_i^\beta} \\ \cdots\cdots\cdots\cdots & \cdot & \cdots\cdots\cdots\cdots\cdots\cdots \\ \displaystyle\sum_{i=1}^n \frac{\log x_i}{x_i^\beta} & \vdots & -\alpha \displaystyle\sum_{i=1}^n \frac{(\log x_i)^2}{x_i^\beta} + \frac{n}{\beta^2} \end{bmatrix} The expected Fisher information matrix of \boldsymbol{\theta} = (\alpha, \beta) based on a single observation is given by \mathbf{I}(\boldsymbol{\theta}) = n \begin{bmatrix} \dfrac{1}{\alpha^2} & \vdots & \dfrac{e^\alpha E_1(\alpha) + 1}{\beta \alpha} \\ \cdots\cdots\cdots\cdots\cdots & \cdot & \cdots\cdots\cdots\cdots\cdots\cdots \\ \dfrac{e^\alpha E_1(\alpha) + 1}{\beta \alpha} & \vdots & \alpha^2 \beta I_{22} + \dfrac{1}{\beta^2} \end{bmatrix}, where E_n(\cdot) represents the exponential integral function defined as E_k(x) = \int_1^\infty \frac{\mathrm{e}^{-xt}}{t^k} \, dt = x^{k-1} \Gamma(1-k, x), and I_{22} = \int_0^1 \left[ \frac{(\log x)^2}{x^\beta} \right] x^{-(\beta+1)} \exp \left[ -\alpha \left( x^{-\beta} - 1 \right) \right] dx.

Text derived from https://en.wikipedia.org/w/index.php?oldid=1376289706 under CC BY-SA 4.0; modifications have been made.

Sources cited in this revision

No external citation was extracted for this revision. That is a statement about the source, not a claim that none exists.