Opinion Creative Commons, CC-BY
The Generalized Sigmoid Almost Power Law Family of Distributions
*Corresponding author: Clement Boateng Ampadu, 31 Carrolton Road, Boston, MA 02132-6303, USA
Received: January 29, 2020; Published: February 19, 2020
DOI: 10.34297/AJBSR.2020.07.001161
Abstract
In [1] we presented a new family of distributions based on the transformation U =xmin ·X, where xmin > 0 and X is a random variable with CDF FX. The new class of distributions appeared to be a good fit to data that are reverse ’J’ shaped in nature. In this short communication we assume X follows the generalized sigmoid generated family of distributions [2], and show a sub-model of the new class of distributions is a good fit to a real-life data set. Consequently, we ask the reader to further investigate some properties and applications of this family.
Keywords and Phrases: Power law probability distributions; Weibull distribution; Generalized Sigmoid generated family of distributions.
Contents
The New Family
At first, we recall the following
Proposition
[1] Let and X is a random variable with CDF FX, then the CDF of U is
Given the cumulative distribution function (CDF) of some baseline distribution as F(x), we defined the generalized sigmoid generated family of distributions with the following CDF
where It follows from
Proposition 1.1:
we have the followingProposition
Suppose the random variable Z has CDF F(z), then, the CDF of the generalized sigmoid almost power law family of distributions is given by
Practical illustration of the new family
We assume Z is a Weibull random variable, so that
where a, b, z > 0. It now follows from Proposition 1.2, that we have the following
Theorem
The CDF of the generalized sigmoid Weibull almost power law family of distributions is given by
Remark
We write if M is a random variable with CDF given by the theorem immediately above. The generalized sigmoid Weibull almost power law family of distributions is practically significant in modelling real-life data as shown below (Figure 1).
Figure 1: The CDF of GSWAPL (1.40849, 165.612, 2.4847, 0.3) fitted to the empirical distribution of the breast cancer data [3].
Remark
Figure 2: The PDF of GSWAPL (1.40849, 165.612, 2.4847, 0.3) fitted to the histogram of the breast cancer data [3].
The PDF of the generalized sigmoid almost power law family of distributions can be obtained by differentiating the CDF (Figure 2).
Concluding Remarks
In this short communication, we introduced a new family of almost power law distributions, and showed a sub-model is practically significant in fitting data in the health sciences. A future interesting problem is to investigate some properties and additional applications of this class of statistical distributions [3].
References
- Clement Boateng Ampadu (2020) A Class of Almost Power Law Distributions. Unpublished.
- Clement Boateng Ampadu (2019) The Generalized Sigmoid Generated Family of Distributions with Illustration to Breast Cancer Data. Unpublished.
- K Jayakumar, M Girish Babu (2017) T-Transmuted X Family of Distributions. Statistica anno LXXVII 77(3): 251-276.