Theory and Applications of the Transmuted Continuous Bernoulli Distribution

Abstract

Modern applied statistics naturally give rise to the continuous Bernoulli distribution (data fitting, deep learning, computer vision, etc). On the mathematical side, it can be viewed as a one-parameter distribution corresponding to a special exponential distribution restricted to the unit interval. As a matter of fact, manageable extensions of this distribution have great potential in the same fields. In this study, we motivate a transmuted version of the continuous Bernoulli distribution with the goal of analyzing proportional data sets. The feature of the created transmuted continuous Bernoulli distribution is an additional parameter that realizes a linear tradeoff between the min and max of two continuous random variables with the continuous Bernoulli distribution. The standard study process is respected: we derive some mathematical properties of the proposed distribution and adopt the maximum likelihood estimation technique in estimating the unknown parameters involved. A Monte Carlo simulation exercise was conducted to examine and confirm the asymptotic behavior of the obtained estimates. In order to show the applicability of the proposed distribution, three proportional data sets are analyzed and the results obtained are compared with competitive distributions. Empirical findings reveal that the transmuted continuous Bernoulli distribution promises more flexibility in fitting proportional data sets than its competitors.