On The Length Biased Quasi-Transmuted Uniform Distribution

International Journal of Engineering Sciences Paradigms and Researches (IJESPR)
Vol. 49, Issue 01, Quarter 01 (January-February-March 2020)

(An Indexed, Peer Reviewed, Referred and Impact Factor Journal)
ISSN (Online): 2319-6564
www.ijesonline.com
On the Length Biased Quasi-Transmuted Uniform

Distribution
Osowole, Oyedeji I.1 and Onyeze, Vitus C.2

1
Reader, Department of Statistics, University of Ibadan, Ibadan, Nigeria
academicprofessor2013@gmail.com
2
Lecturer I, Department of Statistics, Imo State University, Owerri, Nigeria
vitus.c.onyeze@gmail.com
Publishing Date: February 17, 2020
Abstract generalized distributions [1]. Several of these

This study considered another modified Uniform generalized distributions exist in literature. Examples
Distribution, the Length Biased Quasi-Transmuted Uniform include New Extended Burr Type X distribution by
distribution. Some essential properties of the modified [2]; Extended New Generalized Exponential
distribution were derived to show case its flexibility and distribution by [3]; Exponentiated Generalized
tractability. The result indicates that the Length Biased
Quasi-Transmuted Uniform distribution is a competent
Extended Exponential distribution by [4]; Alpha
addition to the existing pool of generalized distributions. Power Transformed Generalized Exponential
Keywords: Length biased quasi-transmuted uniform distribution by [5]; Extended Weighted Exponential
distribution, Moment generating function, Cumulant distribution by [6]; Type I Generalized Exponential
generating function, Hazard function, Order distribution by [7]; Marshall-Olkin Burr Type XII
statistics. distribution by [8]; Scaled Burr Type X distribution
by [9]; Burr Type X distribution by [10]; Extended
Weibull Compound distribution by [11]; Extended
1.0 Introduction Exponential distribution by [12]; Extended Uniform
distribution by [13]; Negative Binomial Marshall-
Development of generalized families of distributions Olkin Rayleigh distribution by [14]; Extended Burr
is as old as Statistics itself. This has always been a Type III distribution by [15]; Quasi-Transmuted
growing research focus of statisticians, applied Uniform distribution by [16] and Quasi-Transmuted
mathematicians, physicists and engineers. The core Complementary Mukherjee-Islam distribution by
aim in this task is to reveal the intrinsic flexibilities [17].
inherent in the extended cum generalized distributing.
Explorations of the properties of these distributions This study however considers the Length Biased
after their development from selected baseline Quasi-Transmuted Uniform Distribution (LBQTUD)
distributions are equally of significant interest. following the previous contributions to knowledge of
the following authors: ([18]; [19]; [20]; [21]; [21] and
Existing distributions may fit badly the dataset under [22]).
consideration and recourse must be made to
IJESPR
www.ijesonline.com
1
www.ijesonline.com
2.0 The Length Biased Quasi-Transmuted Uniform Distribution
The probability density function (pdf) of the Quasi-Transmuted distribution according to Osowole and Ayoola
(2019) is defined as
1
g *( x)  2ax  (1  a), 0  x  1; a  (0, ,1) ……………………………………(1.0)
2
with cumulative distribution function (cdf) defined as
1
G *( x)  ax 2  (1  a) x, 0  x  1; a  (0, ,1) ……………………………………(2.0)
2
A length biased distribution according to Subramanian and Rather (2018) is a weighted distribution fw(x) defined as
w( x) f ( x)
f w ( x)  ,x 0 ……………………………………(3.0)
E[ w( x)]

Where w(x) is non-negative such that E[ w( x)]   w( x) f ( x)dx  

We note that (3.0) becomes a length biased distribution when w(x) is chosen such that w(x) = x and it becomes a
size biased distribution when w(x) is chosen such that w(x) = xc. Combining (1.0) and (3.0), the probability density
function of the length biased quasi transmuted uniform distribution is given as
x[2ax  (1  a)] 1
g LBQTUD ( x)  , 0  x  1; a  (0, ,1)
E[ x] 2
2ax 2  (1  a) x

[(a  3) / 6]
12ax 2  6(1  a) x 1
 , 0  x  1; a  (0, ,1) ……………………………………(4.0)
[(a  3)] 2
(4.0) shall henceforth be referred to in this study as the Length Biased Quasi-Transmuted Uniform Distribution
(LBQTUD). The cumulative distribution function of (4.0) is defined as
x
GLBQTUD ( x)   g LBQTUD (t )dt
0
12at 2  6(1  a)t

x
 dt
0
a3
4ax3  3(1  a) x 2 1
 , 0  x  1; a  (0, ,1) ……………………………………(5.0)
a3 2
IJESPR
www.ijesonline.com
2
www.ijesonline.com
2.1 Some Characterizations of the Length Biased Quasi-Transmuted Uniform

Distribution
2.1.1 Moments
The kth raw moment of a random variable X from the Length Biased Quasi-Transmuted Uniform distribution is
1
k1  E ( X K )   x k g LBQTUD ( x) dx
0
1 1 k

a3 0
x [12ax 2  6(1  a) x]dx
1 12ax k 3 6(1  a) x k  2 1
 [  ]0
a3 k 3 k 2
1 12a(k  2)  6(k  3)(1  a) 

a  3  (k  3)(k  2) 

1  6a(k  1)  6(k  3)  1
   , a  (0, ,1)
a  3  (k  3)(k  2)  2
Specifically,
1 a  2
11  ( X )  mean 
2 5 3 a  3 1 a  3 
 , & for a  0, and 1 respectively
3 7 4 2
Also,
1  9a  15 
21 
a  3  10 
1 19.5 6 1
 , & for a  0, &1 respectively
2 35 10 2
From the
11 and 21 above, the variance of X, V(X )  E ( X   ) 2  2  21  ( 11 ) 2

That is,
IJESPR
www.ijesonline.com
3
www.ijesonline.com
1
 X2  0.056, 0.047 & 0.0375 for a=0, &1 respectively .
2
The variance can be seen to reduce as the value of a is increasing. This is what is expected for a generalized
distribution.
We note also that:
1  4a  6 
31  ( X 3 ) 
a  3  5 
2 8 1 1
= , & for a  0, &1 respectively
5 17.5 2 2 3  ( X   )3  31  321  2 3
= -0.007407,0.007872 and -0.00625 for a  0,
3
Combining 31 and 3 yield the coefficient of skewness, 3  .This implies that
3
1
 3  0.5590, 0.77254 and -0.86066 respectively for a =0, &1 . The skewness can be seen to be
2
reducing as the value of a is increasing. This is not the unexpected pattern for a generalized distribution.
Furthermore,
1  5a  7 
41  ( X 4 ) 
a  3  7 
1 9.5 3 1
= , & for a  0, &1 respectively
3 24.5 7 2
4  ( X   )  4  431  6 2 21  3 4
4 1
1
= 0.007407,0.006247 and 0.00435 for a  0, &1 respectively
2
4
Combining 41 and 4 yield the coefficient of kurtosis, 4  . This implies that
4
1
 4  2.362, 2.828 and 3.093 respectively for a =0, &1 . The coefficient of kurtosis can be seen to be
2
increasing toward 3.0 as the value of a increases from 0 to 1. The value of this coefficient became approximately 3.0
when a reached its maximum value of 1. This indicates that the length biased distribution being considered has the
potentials of a generalized distribution.
IJESPR
www.ijesonline.com
4
www.ijesonline.com
2.1.2 Coefficients of Variation and Dispersion
The coefficients of variance and dispersion for the Length Biased Quasi-Transmuted distribution are
 2
CV  and CD  . Specifically,
E( X ) E( X )
 a 2  2a  5
10(a  3) 2 1
CV  =0.3536,0.3033,0.2582 for a=0, &1
a2 2
 
 a3
 a 2  2a  5 1
and CD  =0.083,0.066,0.05 for a=0, &1
10(a  3)(a  2) 2
The two coefficients above can be seen to be reducing as the value of a is increasing. This is not the unexpected
pattern for a generalized distribution.
2.1.3 Moment Generating Function
The moment generating function (m.g.f.) of a random variable X from the Length Biased Quasi-Transmuted
Uniform distribution is defined as
1
M X (t )  (etX )  e
tX
g LBQTUD ( x)dx
0
1 (tX ) 2 
  1  tX   ...g LBQTUD ( x)dx
0
 2! 
1 
tj
 0 
j 0 j !
X j g LBQTUD ( x)dx

tj 1
 j
j 0 j !
1  t j 12a(j+2)+6(j+3)(1-a)  1
  
a  3 j 0 j !   , a  (0, ,1)
(j+3)(j+2)  2
2.1.4 Characteristic Function
The characteristic function (c. f.) of a random variable X from the Length Biased Quasi-Transmuted Uniform
distribution is defined as
IJESPR
www.ijesonline.com
5
www.ijesonline.com
x (t )  M X (it )
  (eitX )
1
  eitX g LBQTUD ( x)dx
0
1  (it ) j 12a( j  2)  6( j  2)(1  a)  1

  
a  3 J 0 J !  ( j  3)( j  2)  , a  (0, ,1)
 2
2.1.5 Cumulant Generating Function
The cumulant generating function (c. g. f.) of a random variable X from the Length Biased Quasi-Transmuted
Uniform distribution is defined as
K X (t )  In  M X (t ) 
  tj 
 In    1j 
 J o j ! 
 1  t j 12a( j  2)  6( j  2)(1  a)   1
 In      , a  (0, ,1)
 a  3 J 0 j !  ( j  3)( j  2)  2
2.1.6 Hazard Function
The hazard function (h. f.) of a random variable X from the Length Biased Quasi-Transmuted Uniform distribution
is defined as
g LBQTUD ( x)
h( x ) 
1  GLBQTUD ( x)
12ax 2  6(1  a) x 1
 , a  (0, ,1)
4ax  3 x (1  a)  a  3
3 2
2
2.1.7 Reverse Hazard Function
The reserve hazard function (r. h. f.) of a random variable X from the Length Biased Quasi-Transmuted Uniform
distribution is defined as
g LBQTUD ( x)
hr ( x) 
GLBQTUD ( x)
12ax 2  6(1  a) x 1
 , a  (0, ,1)
4ax  3(1  a) x
3 2
2
IJESPR
www.ijesonline.com
6
www.ijesonline.com
2.1.8 Survival Function
The survival function (s.f.) of a random variable X from the Length Biased Quasi-Transmuted Uniform distribution
is defined as
S X ( x)  1  GLBQTUD ( x)
4ax3  3x 2 (1  a)  a  3 1
 , a  (0, ,1)
a3 2
2.1.9 Order Statistics
Suppose X(1), X(2),…, X(n) are the order statistics from the random sample X1, X2, …., Xn from the Length Biased
Quasi-Transmuted Uniform distribution with and as the pdf and cdf, the rth order statistic
where 1  r  n is given as
n! r 1 nr
h( r ) ( x)  g LBQTUD ( x) GLBQTUD ( x)  1  GLBQTUD ( x) 
(r  1)!(n  r )!
By setting r = n and r =1 in the function (h(r)(x)) above, we have the distributions for the largest and lowest order
statistics. For the largest order statistic we have that
n 1
 4ax3  3(1  a) x 2  12ax 2  6(1  a ) x 
h( r  n ) ( x)  n    
 a3   a3 
n 2 n 1 1
  4 ax 3
 3(1  a ) x  12 ax 2
 6(1  a ) x  , a  (0, ,1)
(a  3) n     2
For
The lowest order statistic, we have that

n 1
h( r 1) ( x)  n 1  GLBQTUD ( x)  g LBQTUD ( x)
n 1 n 1
 4ax3  3x 2 (1  a)  a  3  12ax 2  6(1  a) x 
n   
 a3   a3 
n n 1 1
 
n 
4 ax 3
 3 x 2
(1  a )  a  3 
 12ax 2  6(1  a) x  , a  (0, ,1)
(a  3) 2
IJESPR
www.ijesonline.com
7
www.ijesonline.com
2.1.10 Random Number Generation
By the quantile function (i.e the inverse of the cdf), random numbers can be generated for the Length Biased Quasi-
Transmuted Uniform distribution as follows.
1
Let  4ax3  3(1- a) x 2   u so that
a3
1
 4ax3  3(1- a) x 2  - u  0  4ax3  3(1- a) x 2 - u (a  3)  0
a3
where U is a random var iable from the Uniform (0,1). By setting a  0, 4ax3  3(1- a) x 2 - u( a  3)  0
1
becomes x 2  u  x   u ,sin ce U ~ Uniform(0,1). By setting a= , 4ax3  3(1- a) x 2 - u (a  3)  0
2
becomes 2x  1.5 x -3.5u=0. It should be noted that the roots of this equation changes with u. By setting
3 2
a=1, 4ax3  3(1- a) x 2 - u (a  3)  0 becomes 4x 3  4u  0  x   3 u since U~Uniform(0,1). The implication

here is that for the random number generation, a must be set to 0 and 1.
2.1.11 Simulation Study
The table below gives the simulation results from the Length Biased Quasi-Transmuted Uniform distribution at a=0
and 1 for sample sizes 500, 1000, 2000 and 5000 respectively.
Table 1.0a: Simulation from the LBQTUD

a =0 a=1
Sample Summary Sample Summary
Size Statistics Estimate Size Statistics Estimate
500 mean 0.670 500 mean 0.764
variance 0.055 variance 0.033
standard 0.234 standard 0.181
deviation deviation
skewness -0.609 skewness -0.842
kurtosis 2.538 kurtosis 3.071
coefficient of coefficient of
variation 0.349 variation 0.237
dispersion 0.082 dispersion 0.043
1000 mean 0.672 1000 mean 0.757
standard standard 0.191
deviation 0.237 deviation
IJESPR
www.ijesonline.com
8
www.ijesonline.com

2000 mean 0.667 2000 mean 0.745
Table 1.0b: Simulation from the LBQTUD

a=0 a=1
Sample Summary Sample Summary
Size Statistics Estimate Size Statistics Estimate
5000 mean 0.665 5000 mean 0.751
The simulation results in Tables (1.0)a and (1.0)b the expected pattern observable for a generalized
reveal concordance between the simulated results and distribution when compared to the baseline
the expected theoretical results; especially for the distribution and some selected distributions
mean, variance, standard deviation, coefficient of belonging to the same family. The Length Biased
variation and coefficient of dispersion. This further Quasi-Transmuted Uniform Distribution (LBQTUD)
validates the postulated potentials of the Length is a therefore a worthy addition to the existing pool of
Biased Quasi-Transmuted Uniform Distribution as a generalized distributions.
competent addition to the existing families of
generalized distributions.
3.0 Conclusion
References
The specification and characterization of the Length
Biased Quasi-Transmuted Uniform distribution has [1] Subramanian, C. and Rather,
been successfully considered in this study. The A.(2018).Transmuted Generalized Uniform
variance, skewness and kurtosis were found to reduce Distribution, International Journal of Scientific
as the value of a was increasing. Also, the Research in Mathematical and Statistical
coefficients of variation and dispersion decreased as Sciences, 5(3):25-32
the value of a increased as well. This is in line with
IJESPR
www.ijesonline.com
9
www.ijesonline.com
[2] Al-Saiari, A. Y., Baharith, L. A. and Mousa, S. Survival Data. International Journal of Statistics
A. (2017). New Extended Burr Type X and Probability, 2 (1): 63-72
Distribution. Sri Lankan Journal of Applied [14] Jose, K. K. and Sivadas, R. (2015). Negative
Statistics, 17(3): 217-231. Binomial Marshall-Olkin Rayleigh Distribution
[3] Eghwerido, J. T., Zelibe, S. C., Ekuma-Okereke, and its Applications. Economic Quality Control,
E. and Efe-Eyefia, E. (2019). On the Extended 30 (2): 89–98.
New Generalized Exponential Distribution: [15] Al-Saiari, A. Y., Mousa, S. A. and Baharith, L.
Properties and Applications. FUPRE Journal of A. (2016). Marshall-Olkin Extended Burr III
Scientific and Industrial Research, Vol.3 (1):112- Distribution. International Mathematical Forum,
122. 11 (13): 631–642.
[4] Thiago A. N. de Andrade, M. Bourguignon and [16] Osowole, O. I. and Ayoola, F. J. (2019): On the
G. M. Cordeiro. (2016). The Exponentiated Modified Uniform Distribution. Proceedings of
Generalized Extended Exponential Distribution. the 21st iSTEAMS Multidisciplinary
Journal of Data Science 14: 393-414. GoingGlobal Conference, The Council for
[5] Nadarajah, S. and Okorie, I. E. (2017). On the Scientific and Industrial Research-Institute for
Moments of the Alpha Power Transformed Scientific and Technological Information (CSIR-
Generalized Exponential Distribution. Ozone: INSTI) Ghana. 14th-16th November, 2019, Pp.
Science and Engineering, 1-6 81-86.www.isteams.net/goingglobal2019-
[6] Mahdavi, A. and Jabari, L. (2017). An Extended DOIAffix-
Weighted Exponential Distribution. Journal of https://doi.org/10.22624/AIMS/iSTEAMS-
Modern Applied Statistical Methods, 16(1): 296- 2019/V21N1P7
307 [17] Osowole, O. I. (2019). On the Modified
[7] Hamedani G. G., Yousof, H. M., Rasekhi, M., Complementary Mukherjee-Islam Distribution.
Alizadeh, M. and Najibi, M. S. (2018). Type I International Journal of Applied Science and
general exponential class of distributions. Mathematical Theory, 5 (3), 72-79.
Pakistan Journal of Statistics and Operation [18] Das, K.K and Roy, T. D. (2011a). Applicability
Research. 14(1), 39-55. of Length Biased Generated Rayleigh
[8] Al-Saiari, A. Y., Baharith, L. A. and Mousa, S. Distribution. Advances in Applied Science
A. (2014). Marshall-Olkin Extended Burr Type Research, Vol. 2(4): 320-327.
XII Distribution. International Journal of [19] Das, K.K and Roy, T. D. (2011b). On Some
Statistics and Probability, 3(1):78-84. Length-Biased Weighted Weibull Distribution.
[9] Surles, J. G. and Padgett, W. J. (2005). Some Advances in Applied Science Research, Vol.
Properties of a Scaled Burr Type X Distribution. 2(5): 465-475.
Journal of Statistical Planning and Inference, [20] Saghir, A., Tazeem, S., and Ahmad, I. (2016).
128(1): 271–280. The length-biased weighted exponentiated
[10] Raqab, M. Z. and Kundu, D. (2006). Burr Type inverted Weibull distribution. Cogent
X Distribution: Revisited. Journal of Probability Mathematics, 3: 1-18
and Statistical Sciences, 4(2):179–193. [21] Ayesha, A. (2017). Size Biased Lindley
[11] Ghitany, M. E., Al-Hussaini, E. K. and Al- Distribution and Its Properties a Special Case of
Jarallah, R. A. (2005). Marshall-Olkin Extended Weighted Distribution. Applied Mathematics,
Weibull Distribution and its Application to 08(06): 808-819.
Censored Data. Journal of Applied Statistics, [22] Maxwell, O., Oyamakin. S.O., Chukwu, A.U.,
32(10): 1025–1034. Olusola, Y.O. and Kayode, A. A. (2019). New
[12] Srivastava, A. K. and Kumar, V. (2011). Generalization of the Length Biased Exponential
Software Reliability Data Analysis with Distribution with Applications. Journal of
Marshall-Olkin Extended Weibull model using Advances in Applied Mathematics, Vol. 4(2):82-
MCMC method for non-informative set of priors. 88.
International Journal of Computer Applications,
18 (4) : 31–39
[13] Gui, W. (2013). A Marshall-Olkin Power Log-
normal Distribution and its Applications to
IJESPR
www.ijesonline.com
10

On The Length Biased Quasi-Transmuted Uniform Distribution

Uploaded by

Copyright:

Available Formats

On The Length Biased Quasi-Transmuted Uniform Distribution

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

On The Length Biased Quasi-Transmuted Uniform Distribution

Uploaded by

Copyright:

Available Formats

International Journal of Engineering Sciences Paradigms and Researches (IJESPR)

Vol. 49, Issue 01, Quarter 01 (January-February-March 2020)

On the Length Biased Quasi-Transmuted Uniform

Osowole, Oyedeji I.1 and Onyeze, Vitus C.2

Publishing Date: February 17, 2020

Abstract generalized distributions [1]. Several of these

2.0 The Length Biased Quasi-Transmuted Uniform Distribution

12at 2  6(1  a)t

2.1 Some Characterizations of the Length Biased Quasi-Transmuted Uniform

11 and 21 above, the variance of X, V(X )  E ( X   ) 2  2  21  ( 11 ) 2

We note also that:

2.1.2 Coefficients of Variation and Dispersion

2.1.3 Moment Generating Function

2.1.4 Characteristic Function

1  (it ) j 12a( j  2)  6( j  2)(1  a)  1

2.1.5 Cumulant Generating Function

2.1.6 Hazard Function

2.1.7 Reverse Hazard Function

2.1.8 Survival Function

2.1.9 Order Statistics

The lowest order statistic, we have that

2.1.10 Random Number Generation

a=1, 4ax3  3(1- a) x 2 - u (a  3)  0 becomes 4x 3  4u  0  x   3 u since U~Uniform(0,1). The implication

2.1.11 Simulation Study

Table 1.0a: Simulation from the LBQTUD

dispersion 0.083 dispersion 0.049

Table 1.0b: Simulation from the LBQTUD

You might also like