See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/339640892

A New Exponentiated Distribution with Engineering Science Applications

Article in Journal of Probability and Statistics · March 2020

DOI: 10.18576/jsap/090112

CITATIONS READS
7 333

2 authors:

Aafaq A. Rather Subramanian C

Symbiosis International University Annamalai University
40 PUBLICATIONS 202 CITATIONS 29 PUBLICATIONS 165 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Big Data Analytics View project

Papers have been published in Web of science Journals View project

All content following this page was uploaded by Aafaq A. Rather on 16 October 2020.

The user has requested enhancement of the downloaded file.

J. Stat. Appl. Pro. 9, No. 1, 127-137 (2020) 127

Journal of Statistics Applications & Probability

An International Journal

http://dx.doi.org/10.18576/jsap/090112

A New Exponentiated Distribution with Engineering Science

Applications
Aafaq A. Rather*and C. Subramanian

Department of Statistics, Annamalai University, Annamalai Nagar-608002, Tamil Nadu, India.

Received: 6 Mar. 2019, Revised: 8 Apr. 2019, Accepted: 17 June. 2019.

Published online: 1 Mar. 2020.

Abstract: The article addresses various statistical properties and estimation methods for the exponentiated Garima
distribution. The exponentiated Garima distribution has two parameters (scale and shape). Our main focus is on estimation
from the frequentist point of view, yet some statistical and reliability characteristics for the model are derived. We briefly
describe different estimation procedures namely, the maximum likelihood method, order statistics, entropies and reliability
analysis. Finally, the potentiality of the model is analyzed by means of three real lifetime data sets.

Keywords: Exponentiated distribution, Garima distribution, Reliability analysis, Entropies, Order statistics, Maximum
likelihood estimator.

1 Introduction

A new family of distributions, namely the exponentiated exponential distribution was introduced by Gupta et al. see [1].
The family has two parameters: scale and shape, which are similar to the weibull or gamma family. Later Gupta and
Kundu, see [2], studied some properties of the distribution. They observed that many properties of the new family are
similar to those of the weibull or gamma family. Hence the distribution can be used as an alternative to a weibull or gamma
distribution. The two-parameteric gamma and Weibull are the most popular distributions for analyzing any lifetime data.
The gamma distribution has a lot of applications in different fields other than lifetime distributions. The two parameters of
gamma distribution represent the scale and the shape parameter and because of the scale and shape parameter, it has quite a
bit of flexibility to analyze any positive real data. But one major disadvantage of the gamma distribution is that, if the shape
parameter is not an integer, the distribution function or survival function cannot be expressed in a closed form. This makes
gamma distribution little bit unpopular as compared to the Weibull distribution, whose survival function and hazard
function are simple and easy to study. Nowadays exponentiated distributions and their mathematical properties are widely
studied for applied science experimental data sets. Hassan et al. [3], discussed exponentiated Lomax geometric distribution
with its properties and applications. Maradesa Adeleke [4], discussed exponentiated exponential lomax distribution and its
Properties. Nasir et al. see [5], obtained the exponentiated Burr XII power series distribution with properties and its
applications. Nasiru et al. [6], discussed the exponentiated generalized half logistic Burr X distribution and its various
statistical properties. Pal et al. see [7], studied the exponentiated weibull family as an extension of weibull distribution.
Recently, Rather and Subramanian see [8], discussed the exponentiated Mukherjee-Islam distribution which shows more
flexibility than the classical distribution. Rodrigues et al. [9], studied the exponentiated generalized Lindley distribution.
Garima distribution was introduced by Rama Shanker see [10], which is a newly proposed one parametric lifetime
*
Corresponding author e-mail: [email protected]
© 2020 NSP
Natural Sciences Publishing Cor.
128 Aafaq A. Rather, C. Subramanian: A new exponentiated …
model for behavioral and emotional science. The usefulness and importance of the proposed distribution in lifetime data
were greater as compared to lindley and the exponential distribution. Furthermore, Garima distribution has been modified
and generalized by so many researchers. Rama Shanker [11], discussed the discrete Poisson-Garima distribution. Also
Shanker and Shukla see [12], discussed Zero-Truncated Poisson-Garima distribution and its applications. In this paper, we
consider a two-parameter exponentiated Garima distribution and study some of its properties. The two parameters of
exponentiated Garima distribution represent the shape and the scale parameters. It also has the increasing or decreasing
failure rate depending of the shape parameter. The density function varies significantly depending of the shape parameter
(see Fig.1). Finally, the three real lifetime data sets have been analyzed, the results are compared over Garima and
Exponential distribution and has been found that exponentiated garima distribution provides better fit than Garima and
exponential distribution.

2 Exponentiated Garima (EG) Distributions

The probability density function (pdf) of Garima distribution is given by

q
g ( x) = (1 + q + qx)e -qx ; x > 0, q > 0 (1)
q +2
and its cumulative distribution function (cdf) is given by

æ qx ö -qx
G ( x ) = 1 - ç1 + ÷e ; x > 0, q > 0 (2)
è q +2ø

A random variable X is said to have an exponentiated distribution if its cumulative distribution function (cdf) is given by

Fa ( x) = (G( x)) ,
a
x Î R1 , a > 0 (3)

Then X is said to have an exponential distribution.

The probability density function (pdf) of X is given by

f a ( x) = a (G( x))
a -1
g ( x) (4)

On substituting (2) in (3), we will get the cumulative density function of exponentiated Garima distribution
a
æ æ qx ö -qx ö
Fa ( x) = çç1 - ç1 + ÷ e ÷÷ ; x > 0,q > 0, a > 0 (5)
è è q + 2ø ø
and the probability density function of EG distribution can be obtained as

a -1
aq (1 + q + qx)e -qx æ æ qx ö -qx ö
f a ( x) = çç1 - ç1 + ÷ e ÷÷ (6)
q +2 è è q + 2ø ø
3 Reliability Analysis
In this section, we will discuss about the survival function, failure rate, reverse hazard rate and Mills ratio of the EG
distribution.
The survival function or the reliability function of the exponentiated Garima distribution is given by

a
æ æ qx ö -qx ö
S ( x) = 1 - çç1 - ç1 + ÷ e ÷÷
è è q + 2ø ø

© 2020 NSP
Natural Sciences Publishing Cor.
J. Stat. Appl. Pro. 9, No. 1, 127- 138 (2020) / http://www.naturalspublishing.com/Journals.asp 129

The hazard function is also known as the hazard rate, instantaneous failure rate or force of mortality and is given by
a -1
æ aq (1 + q + qx)e -qx æ æ qx ö -qx ö ö÷
ç çç1 - ç1 + ÷ e ÷÷
ç q +2 è è q + 2ø ø ÷
h( x ) = ç a ÷
ç æ æ qx ö -qx ö ÷
ç 1 - çç1 - ç1 + ÷ e ÷÷ ÷
è è è q + 2ø ø ø
The reverse hazard rate is given by

aq (1 + q + qx)e -qx
hr ( x ) =
(q + 2) - (q + qx + 2)e -qx
and the Mills ratio of the EG distribution is
(q + 2) - (q + qx + 2)e -qx
Mills Ratio =
aq (1 + q + qx)e -qx

Fig 1: pdf plot of exponentiontiated Gama Distribution. Fig 2:Cdf plot of Exponentiated Garima distribution.

© 2020 NSP
Natural Sciences Publishing Cor.
130 Aafaq A. Rather, C. Subramanian: A new exponentiated …

Fig 3: Survival function curves of EG distribution.

4 Statistical Properties

In this section, we will discuss the different structural properties of the proposed exponentiated Garima distribution.

4.1 Moments

Suppose X is a random variable following exponentiated Garima distribution with parameters α and θ, then the rth order
moment E(xr) for a given probability distribution is given by
¥
E ( X ) = µ r = ò x r f a ( x)dx
r '

0
¥ a -1
aq (1 + q + qx)e -qx æ æ qx ö -qx ö
= òx r
çç1 - ç1 + ÷ e ÷÷ dx
0
q +2 è è q + 2ø ø

a -1
æ ö
¥
ç ÷
aq -qx ç æ qx ö -qx ÷
q + 2 ò0
= x (1 + q + qx)e ç1 - ç1 +
r
÷e ÷ dx ( 7)
è q + 2ø
çç ÷÷
è ø

a -1 i
æ æ qx ö -qx ö ¥
æ a - 1öìæ qx ö -qx ü
Using binomial expansion of çç1 - ç1 + ÷ e ÷÷ = å çç ÷÷íç1 + ÷e ý (-1) , equation (7) will
i

è è q + 2ø ø i =0 è i øîè q + 2 ø þ
become

i æ a - 1ö
¥ i
aq ¥ æ qx ö
E( X ) = r
å
q + 2 i =o
(-1) çç ÷÷ ò x r (1 + q + qx)e -qx (1+i ) ç1 +
è q + 2ø
÷ dx (8)
è i ø0
Again using binomial expansion in equation (8), we get
k +1 ¥
¥
æa - 1ö æ i ö æ q ö
¥
E ( X ) = a åå (-1) çç
r
÷÷ çç ÷÷ çç
i
÷÷ òx
r +k
(1 + q + qx)e -qx (1+i ) dx
i =0 k =0 è i ø è k ø è (q + 2 ø 0

After simplification, we obtain

k +1
¥ ¥
æ a - 1ö æ i ö æ q ö
E ( X ) = a åå (-1) çç
r i
÷÷ çç ÷÷ çç ÷÷
i =0 k =0 è i ø è k ø è (q + 2 ø
q G(r + k + 1)
´ ((1 + q )(1 + i) + (r + k + 1)) (9)
(q (1 + i))r +k +2

Since equation (9) is a convergent series for all r ³ 0 , therefore all the moments exist.

Therefore

© 2020 NSP
Natural Sciences Publishing Cor.
J. Stat. Appl. Pro. 9, No. 1, 127- 138 (2020) / http://www.naturalspublishing.com/Journals.asp 131

k +1
¥
æa - 1ö æ i ö æ q ö
¥
q G(k + 2)
E ( X ) = a åå (-1) çç ÷÷ çç ÷÷ çç
i
÷÷ ((1 + q )(1 + i) + (k + 2))
i =0 k =0 è i ø è k ø è (q + 2 ø (q (1 + i))k +3
and
k +1
¥
æa - 1ö æ i ö æ q ö
¥
q G(k + 3)
E ( X ) = a åå (-1) çç
2
÷÷ çç ÷÷ ççi
÷÷ ((1 + q )(1 + i) + (k + 3))
i =0 k =0 è i ø è k ø è (q + 2 ø (q (1 + i))k +4
Therefore, the variance of X can be obtained as

V ( X ) = E ( X 2 ) - (E ( X ) )
2

4.2 Harmonic Mean

The harmonic mean for the EG distribution can be obtained as

æ1ö ¥1
H.M = E ç ÷ = ò f a ( x)dx
è xø 0 x

¥ a -1
1 aq (1 + q + qx)e -qx æ æ qx ö -qx ö
=ò çç1 - ç1 + ÷ e ÷÷ dx
0
x q +2 è è q + 2ø ø

a -1
aq ¥ 1 æ æ qx ö -qx ö
= ò
q +2 0 x
(1 + q + qx)e -qx çç1 - ç1 +
è è q + 2ø
÷ e ÷÷
ø
dx

After simplification, we obtain

k +1
¥
æ a - 1öæ i öæ
¥
1 ö
H .M . = a åå (-1) çç ÷÷çç ÷÷çç
i
÷÷ (q (1 + q )(1 + i) + qk )Gk
i =0 k =0 è i k
øè øè (q + 2)(1 + i ) ø

4.3 Moment Generating Function and Characteristic Function

Let X have exponentiated Garima distribution, then the moment generating function of X is obtained as
¥
M X (t ) = E (e ) = ò e tx fa ( x)dx
tx

Using Taylor’s series

¥
æ (tx) 2 ö
M X (t ) = ò çç1 + tx + + × × × × ÷÷ f a ( x)dx
0è
2! ø

© 2020 NSP
Natural Sciences Publishing Cor.
132 Aafaq A. Rather, C. Subramanian: A new exponentiated …

¥ ¥
tj j
= òå x f a ( x)dx
0 j =0 j!

¥
tj
= å µj'
j =0 j!

k +1
¥ ¥ ¥
æ a - 1ö æ i ö æ q ö
M X (t ) = a ååå (-1) çç i
÷÷ çç ÷÷ çç ÷÷
i =0 j =0 k =o è i ø è k ø è (q + 2 ø
t j q G( j + k + 1)
´ ((1 + q )(1 + i) + ( j + k + 1))
j! (q (1 + i) ) j + k + 2

Similarly, the characteristic function of EG distribution is given by

k +1
¥ ¥
æ a - 1ö æ i ö æ q ö
¥
j X (t ) = a ååå (-1) çç i
÷÷ çç ÷÷ çç ÷÷
i =0 j =0 k =o è i k
øè øè (q + 2 ø
(mt ) q G( j + k + 1)
j
´ ((1 + q )(1 + i) + ( j + k + 1))
j! (q (1 + i) ) j + k + 2

5 Information Measures of EG Distribution

5.1 Renyi Entropy

Entropies quantify the diversity, uncertainty, or randomness of a system. The Renyi entropy is named after Alfred Renyi in
the context of fractual dimension estimation, the Renyi entropy forms the basis of the concept of generalized dimensions.
The Renyi entropy is important in ecology and statistics as index of diversity. The Renyi entropy is also important in
quantum information, where it can be used as a measure of entanglement. For a given probability distribution, Renyi
entropy is given by
1 æ¥ b ö
e (b ) = logçç ò f ( x)dx ÷÷
1- b è0 ø
where, b > 0 and b ¹ 1
é aq ¥ ìï qx ö -qx ö
b (a -1)
üïù
1 bæ æ
òí
- bq x
= log ê e (1 + q + qx ) ç
ç1 - ç1 + ÷e ÷
÷ ýú dx
1- b êëq + 2 0 ïî è è q + 2ø ø ïþúû

After simplification, we get

1 éæ aq ö b ¥ ¥ ¥ b - j æ b (a - 1) öæ i öæ k öæ q ö
k ¥
ù
÷ ååå (-1) (1 + q ) çç ò0
-q ( b + i ) x ( j + k +1) -1
e( b ) = log êç i
÷÷çç ÷÷çç ÷÷ç ÷ q
j
e x dx ú
1- b ëêè q + 2 ø i =0 k =0 j =0 è i øè k øè j øè q + 2 ø ûú

© 2020 NSP
Natural Sciences Publishing Cor.
J. Stat. Appl. Pro. 9, No. 1, 127- 138 (2020) / http://www.naturalspublishing.com/Journals.asp 133

1 é ¥ ¥ ¥
æ b (a - 1) ö æ i ö æ k ö q b -1 G( j + k + 1) ù
e( b ) = log êa b ååå (-1) i (1 + q ) b - j çç ÷÷ çç ÷÷ çç ÷÷ ú
1- b ë i =0 k =0 j =0 è i ø è k ø è j ø (q + 2)
b +k
( b + i) j + k +1 û

5.2 Tsallis Entropy

A generalization of Boltzmann-Gibbs (B-G) statistical mechanics initiated by Tsallis has gained a great deal of attention.
This generalization of B-G statistics was proposed firstly by introducing the mathematical expression of Tsallis entropy,
see [13], for a continuous random variable, it is defined as follows

1 æ ö
¥
Sl = ç1 - ò f l ( x)dx ÷
l - 1 çè 0 ÷
ø
1 é üïù
l (a -1)
aq ¥ ìï -lqx læ æ qx ö -qx ö
l - 1 ê q + 2 ò0 ïî
= ê1- íe (1 + q + qx) çç1 - ç1 + ÷e ÷÷ ýú dx
è è q + 2 ø ø ïþúû
ë
After simplification, we get

l
1 é æ aq ö ¥ ¥ ¥ l - j æ l (a - 1) öæ i öæ k öæ q
ù
k ¥
ö
÷ ååå (-1) (1 + q ) çç ò0
-q ( l + i ) x ( j + k +1) -1
Sl = ê1 - ç ÷ç ÷ç ÷ç ÷ q
i j
e x dx ú
l - 1 ëê è q + 2 ø i =0 k =0 j =0 è i ÷øçè k ÷øçè j ÷øè q + 2 ø ûú
1 é l - j æ l (a - 1) ö æ i ö æ k ö q l -1 G( j + k + 1) ù
¥ ¥ ¥

ê1 - a ååå (-1) (1 + q ) çç
l
Sl = i
÷ç ÷ç ÷ ú
l -1 ë i =0 k =0 j =0 è i ÷ø çè k ÷ø çè j ÷ø (q + 2) l + k (l + i) j + k +1 û

6 Bonferroni And Lorenz Curves

The Bonferroni and the Lorenz curves are not only used in economics in order to study the income and poverty. It has also
various applications in other fields like reliability, medicine, insurance and demography. The Bonferroni and Lorenz curves
are given by
q
1
pµ1 ' ò0
B( p) = xf ( x)dx

and
q
1
µ1 ' ò0
L( p) = PB( p) = xf ( x)dx

and q = F -1 ( p )
k +1
¥
æa - 1ö æ i ö æ q ö
¥
q G(k + 2)
Where µ1 = a åå (-1) çç
'
÷÷ çç ÷÷ çç
i
÷÷ ((1 + q )(1 + i) + (k + 2))
i =0 k =0 è i ø è k ø è (q + 2 ø (q (1 + i))k +3
q a -1
1 aq (1 + q + qx)e -qx æ æ qx ö -qx ö
' ò
B( p ) = x çç1 - ç1 + ÷ e ÷÷ dx
pµ1 0 q +2 è è q + 2ø ø

After simplification, we get

© 2020 NSP
Natural Sciences Publishing Cor.
134 Aafaq A. Rather, C. Subramanian: A new exponentiated …

1 ¥ ¥ æ a - 1ö æ (1 + q )(1 + i)g ((k + 2),q (1 + i)q ) + g ((k + 3),q (1 + i)q ) ö

B( p) = åå ç ÷ ç
p i =0 i =0 çè i ÷ø çè ((1 + q )(1 + i) + (k + 2))G(k + 2)
÷÷
ø
and
æa - 1ö æ (1 + q )(1 + i)g ((k + 2),q (1 + i)q ) + g ((k + 3),q (1 + i)q ) ö
¥ ¥
L( p) = PB( p) = åå çç ÷ç ÷÷
i =0 i =0 è i ÷ø çè ((1 + q )(1 + i) + (k + 2))G(k + 2) ø
7 Maximum Likelihood Estimation
In this section, we will discuss the maximum likelihood estimators of the parameters of EG distribution.
Let X1, X2, ..., Xn be the random sample of size n from the EG distribution, then the likelihood function can be written as
a -1
(aq )n næ -qx æ æ qx ö -qx ö ö÷
L(a ,q ) =
(q + 2)n
Õ ç
ç
i =1 è
(1 + q + qx)e ç1 - ç1 +
è è q + 2ø
÷e ÷ ÷
ø ø
The log likelihood function is
n
log L(a , q ) = n log a + n log q - n log(q + 2) + å log(1 + q + qxi )
i =1
n n
æ æ qxi ö -qxi ö
- q å xi +(a - 1)å logçç1 - ç1 + ÷e ÷÷ (10)
i =1 i =1 è è q + 2 ø ø
The maximum likelihood estimates of α, θ which maximize (10), must satisfy the normal equations given by

¶ log L n n æ æ qx ö ö
= + å logçç1 - ç1 + i ÷e -qxi ÷÷ = 0
¶a a i =1 è è q + 2 ø ø
! n
Þa = n
ìæ qxi ö -qxi ü
å logíç1 + ÷e + 1ý
i =1 îè q + 2 ø þ
and
¶ log L n n n n
æ 1 + xi ö æ æ qx ö ö
= - - å xi + å çç ÷÷ + (a - 1)y çç1 - ç1 + i ÷e -qxi ÷÷ = 0
¶q q q + 2 i =1 i =1 è 1 + q + qxi ø è è q + 2ø ø

Where ψ(.) is the digamma function.

At this point, it is important to highlight that the analytical solution of the above system of non-linear equation is unknown.
Algebraically it is very difficult to solve the complicated form of likelihood system of nonlinear equations. Therefore we
use R and wolfram mathematics for estimating the required parameters.

8 Order Statistics
Order statistics has been widely used in the field of reliability and life testing. There is also an extensive role of order
statistics in several aspects of statistical inference. Let X(1), X(2), ...,X(n) be the order statistics of a random sample X1, X2, ...,
Xn drawn from the continuous population with probability density function fx(x) and cumulative density function with
Fx(x), then the pdf of rth order statistics X(r) can be written as

© 2020 NSP
Natural Sciences Publishing Cor.
J. Stat. Appl. Pro. 9, No. 1, 127- 138 (2020) / http://www.naturalspublishing.com/Journals.asp 135

n!
f X ( x) [FX ( x)] [1 - FX ( x)]
r -1 n-r
f X ( r ) ( x) = (11)
(r - 1)!(n - r )!

Substitute the values of (5) and (6) in equation (11), we will get the pdf of rth order statistics X(r) for exponentiated Garima
distribution and is given by
a -1
n! aq (1 + q + qx)e -qx æ æ qx ö -qx ö
f X ( r ) ( x) = çç1 - ç1 + ÷ e ÷÷
(r - 1)!(n - 1)! q +2 è è q + 2ø ø
a ( r -1) a n-r
qx ö -qx ö æ æ æ ö
æ æ
´çç1 - ç1 + ÷ e ÷÷ ç1 - ç1 - ç1 + qx ö÷ e -qx ö÷ ÷
è è q + 2ø ø ç çè è q + 2 ø ÷ ÷
ø ø
è

The probability density function of higher order statistics X(n) can be obtained as

a -1 a ( n -1)
aq (1 + q + qx)e -qx æ æ qx ö -qx ö æ æ qx ö -qx ö
f X ( n ) ( x) = n çç1 - ç1 + ÷ e ÷÷ çç1 - ç1 + ÷ e ÷÷
q +2 è è q + 2ø ø è è q + 2ø ø

Similarly, the pdf of 1st order statistic X(1)can be obtained as

a -1 a n -1
aq (1 + q + qx)e -qx qx ö -qx ö æ æ æ ö
f X (1) ( x) = n
æ æ
çç1 - ç1 + ÷ e ÷÷ ç1 - ç1 - ç1 + qx ö÷ e -qx ö÷ ÷
q +2 è è q + 2ø ø ç çè è q + 2 ø ÷ ÷
ø ø
è
9 Data Analysis
In this section, we have used three real lifetime data sets in exponentiated Garima distribution and the model has been
compared with Garima and Exponential distribution.

Data set 1: The data set is reported by (see [14]), which is related with strength data of window glass of the aircraft of 31
windows. The data are

18.83 20.80 21.657 23.03 23.23 24.05 24.321 25.50

25.52 25.80 26.69 26.77 26.78 27.05 27.67 29.90
31.11 33.20 33.73 33.76 33.89 34.76 35.75 35.91
36.98 37.08 37.09 39.58 44.045 45.29 45.431
Data set 2: The data set is reported by (see [15]). The following data are related to failure times of the air conditioning
system of an airplane. The data are given below:
23, 261, 87, 7, 120, 14, 62, 47, 225, 71, 246, 21, 42,
20, 5, 12, 120, 11, 3, 14, 71, 11, 14, 11, 16, 90,
1, 16, 52, 95

Data set 3: The data set represents the tensile strength, measured in GPa, of 69 carbon fibers tested under tension at gauge
lengths of 20mm which were originally reported by (see [16]), the data are given as:
1.312, 1.314, 1.479, 1.552, 1.700, 1.803,
1.861, 1.865, 1.944, 1.958, 1.966, 1.997,
2.006, 2.021, 2.027, 2.055, 2.063, 2.098,
2.14, 2.179, 2.224, 2.240, 2.253, 2.270,

© 2020 NSP
Natural Sciences Publishing Cor.
136 Aafaq A. Rather, C. Subramanian: A new exponentiated …
2.272, 2.274, 2.301, 2.301, 2.359, 2.382,
2.382, 2.426, 2.434, 2.435, 2.478, 2.490,
2.511 2.514, 2.535, 2.554, 2.566, 2.57,
2.586, 2.629, 2.633, 2.642, 2.648, 2.684,
2.697, 2.726, 2.770, 2.773, 2.800, 2.809,
2.818, 2.821, 2.848, 2.88, 2.954, 3.012,
3.067, 3.084, 3.090, 3.096, 3.128, 3.233,
3.433, 3.585, 3.585.

In order to compare the exponentiated Garima distribution with Garima and Exponential distribution, we consider
the criteria like Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), Akaike Information Criterion
Corrected (AICC) and -2 logL. The better distribution is which corresponds to lesser values of AIC, BIC, AICC and – 2 log
L. For calculating AIC, BIC, AICC and -2 logL can be evaluated by using the formulas as follows:

2k (k + 1)
AIC = 2K - 2logL, BIC = klogn - 2logL, AICC = AIC +
(n - k - 1)
Where k is the number of parameters, n is the sample size and -2 logL is the maximized value of log likelihood function and
they are showed in Table 1.
Table 1: Fitted distributions of the three data sets and criteria for comparison.
Data Distribution MLE S.E -2 logL AIC BIC AICC
Sets
Exponentiated 𝜃"=0.19028545 𝜃"=0.02522012 208.2141 212.2141 215.082 212.642671
Garima 𝛼$=54.28826739 𝛼$=31.57442845
1 Garima 𝜃"=0.051574492 𝜃"=0.007885046 267.278 269.278 270.712 269.415931

Exponential 𝜃"=0.032461087 𝜃"=0.005824644 274.5289 276.5289 277.9629 276.666831

Exponentiated 𝜃"=0.019286347 𝜃"=0.004468638 305.7035 309.7035 311.5059 309.7035

Garima 𝛼$=0.689518572 𝛼$=0.153184409
2
Garima 𝜃"=0.024054339 𝜃"=0.003909228 308.7266 310.7266 312.1277 310.869457

Exponential 𝜃"=0.016793310 𝜃"=0.003055108 315.2594 317.2594 318.6606 317.402257

Exponentiated 𝜃"=2.2829702 𝜃"=0.1867621 108.4583 112.4583 116.9265 112.640118

Garima 𝛼$=70.0895927 𝛼$=26.0879240
3
Garima 𝜃"=0.58724915 𝜃"=0.05988194 251.0992 253.0992 255.3333 253.158901
Exponential 𝜃"=0.40794208 𝜃"=0.04911016 261.7352 263.7352 265.9693 263.794901

2k (k + 1)
Where, AIC = 2k - 2 log L, BIC = k log n - 2 log L and AICC = AIC +
(n - k - 1)

It can be easily seen from Table 1, that the exponentiated Garima distribution has the lesser AIC, BIC, AICC and -2 logL
values as compared to Garima and Exponential distributions. Hence, we can conclude that the exponentiated Garima
distribution leads to better fit than the Garima and Exponential distributions.

10 Conclusions
In the present study, we have introduced a new generalization of the Garima distribution namely as exponentiated Garima
distribution with two parameters (scale and shape). The subject distribution is generated by using the exponentiated

© 2020 NSP
Natural Sciences Publishing Cor.
 J. Stat. Appl. Pro. 9, No. 1, 127- 138 (2020) / http://www.naturalspublishing.com/Journals.asp 137

technique and the parameters have been obtained by using maximum likelihood estimator. Some mathematical properties
along with reliability measures are discussed. The new distribution with its applications in real life-time data has been
demonstrated. Finally, the results of three real lifetime data sets have been compared over Garima and exponential
distributions and have been found that exponentiated Garima distribution provides better fit than Garima and exponential
distributions.
Acknowledgement
The authors thank the referees for their constructive suggestions, which have immensely helped the presentation of the
paper and also thank the editor for their valuable suggestions.
References
[1] R. C. Gupta, P. L. Gupta and R. D. Gupta, Modeling failure time data by Lehman alternatives. Communications in Statistics, Theory
and Methods, 27, 887–904 (1998).
[2] R.D. Gupta and Kundu, Exponentiated exponential family: an alternative to gamma and Weibull, Biometrical Journal., 43(1), 117-
130 (2001).
[3] A. S. Hassan and M. A. Abdelghafar, Exponentiated Lomax Geometric Distribution: Properties and Applications, Pakistan Journal
of Statistics and Operation Research, 13(3) (2017).
[4] Maradesa Adeleke, Exponentiated Exponential Lomax Distribution and its Properties, Mathematical Theory and Modeling.
doi:10.7176/mtm/9-1-01 (2019).
[5] A. Nasir, H. Yousof, F. Jamal, and M. Korkmaz, The Exponentiated Burr XII Power Series Distribution: Properties and
Applications. Stats, 2(1), 15-31 (2018).
[6] S. Nasiru, P. N. Mwita and O. Ngesa, Exponentiated Generalized Half Logistic Burr X Distribution, Advances and Applications in
Statistics., 52(3), 145-169 (2018).
[7] M. Pal, M. M. Ali and J. Woo, Exponentiated weibull distribution, Statistica, auno, LXVI.,2, 139-147 (2006).
[8] A. A. Rather and C. Subramanian, Exponentiated Mukherjee-Islam Distribution, Journal of Statistics Applications & Probability,
7(2), 357-361 (2018).
[9] J. Rodrigues, A. Percontini and G. Hamedani, The Exponentiated Generalized Lindley Distribution, Asian Research Journal of
Mathematics, 5(3), 1-14 (2017).
[10] R. Shanker, Garima distribution and its Application to Model Behavioral Science Data, BiomBiostat Int J 4(7), 00116 (2016).
[11] R. Shanker, The Discrete Poisson-Garima distribution, Biom Biostat Int J., 5(2) (2017).
[12] S. Rama, K. S. Kamlesh, Zero-Truncated Poisson-Garima Distribution and its Applications, Biostat Biometrics Open Acc J., 3(1), 1-
5 (2017).
[13] C. Tsallis, Possible generalization of boltzmann-gibbs statistics, Journal of Statistical Physics, 52, 479-487 (1998).
[14] E. J. Fuller, S. Frieman, J. Quinn, G. Quinn, and W. Carter, Fracture mechanics approach to the design of glass aircraft windows: A
case study, SPIE Proc., 2286, 419-430 (1994).
[15] H. Linhart and W. Zucchini, Model Selection, Wiley, NewYork, (1986).
[16] M. G. Bader and A. M. Priest, Statistical aspects of fiber and bundle strength in hybrid composites, In; hayashi T, Kawata K.
Umekawa S (Eds.), Progress in Science in Engineering Composites, ICCM-IV Tokyo., 1129 -1136 (1982).

Aafaq A. Rather, is a Research Scientist at the Department of Statistics, Annamalai University, Tamil Nadu, India.
He did his M.Sc. in Statistics from University of Kashmir, Srinagar with Distinction and M.Phil from Agra University,
Agra, India and topped the academic honors and has been a Gold Medalist. His research interests are in the areas of
Generalized Probability distributions, Bio-Statistics and Stochastic Modeling. His research interests are in the areas of
Generalized Probability distributions, Bio-Statistics and Stochastic Modeling. He has published research articles in
reputed both international and national journals.

C. Subramanian did his M.Sc and M.Phil degrees in Statistics from Department of Statistics, Bharathiar University,
Coimbatore, India in 1989 and 1990 respectively. He received his Ph.D degree in Statistics from Department of
Statistics, Annamalai University, Chidambaram, India in 2003. At present he is Assistant Professor in the Department
of Statistics, Annamalai University, Chidambaram, India. He has published work in international and national journals
of repute. His current research areas are in the Stochastic Modeling, Bio-Statistics and Generalized Distributions.

© 2020 NSP
Natural Sciences Publishing Cor.

View publication stats