American Open Journal of Statistics, 2011, 1, 87-92

doi:10.4236/ojs.2011.12010 Published Online July 2011 (http://www.SciRP.org/journal/ojs)

Copyright 2011 SciRes. OJS
Sequential Test of Fuzzy Hypotheses
Mohammad Ghasem Akbari
Department of Statistics, Faculty of Sciences, University of Birjand, Southern Khorasan, Iran
E-mail: g_z_akbari@yahoo.com
Received May 22, 2011; revised June 10, 2011; accepted June 17, 2011
Abstract

In testing statistical hypotheses, as in other statistical problems, we may be confronted with fuzzy concepts.
This paper deals with the problem of testing hypotheses, when the hypotheses are fuzzy and the data are
crisp. We first give new definitions for notion of mass (density) probability function with fuzzy parameter,
probability of type I and type II errors and then state and prove the sequential probability ratio test, on the
basis of these new errors, for testing fuzzy hypotheses. Numerical examples are also provided to illustrate the
approach.

Keywords: Canonical Fuzzy Number, Fuzzy Hypotheses, Type I and II Error Sizes, Sequential Probability
Ratio Test
1. Introduction

Statistical analysis, in traditional form, is based on cri-
spness of data, random variable, point estimation, hypo-
theses, parameter and so on. As there are many different
situations in which the above mentioned concepts are
imprecise. On the other hand, the theory of fuzzy sets is a
well known tool for formulation and analysis of impre-
cise and subjective concepts. Therefore the sequential
probability ratio test with fuzzy hypotheses can be
important. The problem of statistical inference in fuzzy
environments are developed in different approaches.
Delgado et al. [1] consider the problem of fuzzy hypo-
theses testing with crisp data. Arnold [2,3] presents an
approach to test fuzzily formulated hypotheses, in which
he considered fuzzy constraints on the type I and II
errors. Holena [4] considers a fuzzy generalization of a
sophisticated approach to exploratory data analysis, the
general unary hypotheses automaton. Holena [5] presents
a principally different approach and motivates by the
observational logic and its success in automated know-
ledge discovery. Neyman-pearson lemma for fuzzy hy-
potheses testing and Neyman-pearson lemma for fuzzy
hypotheses testing with vague data is given by Taheri et
al. and Torabi et al. [6,7]. Filzmoser and Viertl [8]
present an approach for statistical testing at the basis of
fuzzy values by introducing the fuzzy p-value. Some
methods of statistical inference with fuzzy data, are
reviewed by Viertl [9]. Buckley [10,11] studies the
problems of statistical inference in fuzzy environment.
Thompson and Geyer [12] proposed the Fuzzy p-values
in latent variable problems. Taheri and Arefi [13] exhibit
an approach for testing fuzzy hypotheses based on fuzzy
test statistics. Parchami et al. [14] consider the problem
of testing hypotheses, when the hypotheses are fuzzy and
the data are crisp. they first introduce the notion of fuzzy
p-value, by applying the extension principle and then
present an approach for testing fuzzy hypotheses by
comparing a fuzzy p-value and a fuzzy significance level,
based on a comparison of two fuzzy sets.
In present work, we first define a new approach for
obtaining the probability (density) function, when the
random variable is crisp and the parameter of interest is
imprecise (fuzzy). Also, the type I and type II errors are
introduced based on fuzzy hypotheses. Then, the
sequential probability ratio test (SPRT) is defined and
extended based on such hypotheses.
We organize the matter in the following way:
In section 2 we describe some basic concepts of fuzzy
hypotheses, density (Mass) probability function with
fuzzy parameter and necessary definitions. In section 3
we come up sequential probability ratio test based on
fuzzy hypotheses. In section 4 the previous definitions
and the sequential probability ratio test will be illustrated
by examples.

2. Preliminaries

In this section we describe fuzzy hypotheses, density
(Mass) probability function with fuzzy parameter and
M. G. AKBARI 88


)
necessary definitions.
Let be a probability space, a random
variable (RV)
( , , P O
X is a measurable function from
to , where
X
is the probability
measure induced by
( ) P , , O ( , , )
X
P P
X and is called the distribution of
the RV X , i.e.,
( ) ( ) = = d
X
X A
P A P X A P A
e
e
}
. e
If is dominated by a
X
P o finite measure u , i.e.
X
P u then by the Radon-Nikodym theorem
(Billingsley, [15]), we have
( ) ( ) ( ) = d
X
X A
P A f x x A u u
e
e
}
,
where
( )
f x u is the Radon-Nikodym derivative of
X
with respect to P u and is called the probability
density function of X with respect to u . In a
statistical context, the measure u is usually a counting
measure or a Lebesgue measure, hence is ( )
X
P A
( )
=
x A
P X x u
e

or
( )
d
A
f x x u
}
, respectively.

2.1. Canonical Fuzzy Numbers

Let
( ) {
= f x u u O
}
> 0 be the support or sample
space of , then a fuzzy subset O u

of
X
is defined
by its membership function
S
| | : O 0.1
u

}
. We denote
by ( ) {
x
o u
: u u o = >

the o cut set of u

and
0
u

is the closure of the set ( ) x
u
u
{ }
: 0 >

1)
, and
u

is called a normal fuzzy set if there exists

u eO such that ; ( ) 1 x
u
=

2) u

is called a convex fuzzy set if

( ) ( ) ( ) ( ) ( )
1 min , x y x
u u
+ >

|
y
u

for all
| 0,1 e ;
3) u

is called a fuzzy number if u

is a normal
convex fuzzy set and its o cut sets, are bounded
0 o = ;
4) u

is called a closed fuzzy number if u

is a
fuzzy number and its membership function
u

is upper
semicontinuous;
5) u

is called a bounded fuzzy number if u

is a
fuzzy number and the support of its membership function
u

is compact.
If u

is a closed and bounded fuzzy number with

{ } :
L
= inf
o o
u u u u e and { = sup :
U
o o
} u u u u e and its
membership function be strictly increasing on the
interval
1
,
L L
o
u u (

and strictly decreasing on the interval
1 o
,
U U
u u (

, then u

is called a canonical fuzzy number

(Klir and Yuan, [16]).
The fuzzy canonical numbers (such as triangular or
trapezoidal fuzzy numbers) are very realistic in fuzzy set
theory, so we use this numbers for our goal.
2.2. Fuzzy Hypotheses

We define some models, as fuzzy sets of real numbers,
for modeling the extended versions of the simple, the
one-sided, and the two-sided ordinary (crisp) hypotheses
to the fuzzy ones.
Testing statistical hypothesis is a main branch of
statistical inference. Typically, a statistical hypothesis is
an assertion about the probability distribution of one or
more random variable(s). Traditionally, all statisticians
assume the hypothesis for which we wish provide a test
are well-defined. This limitation, sometimes, force the
statistician to make decision procedure in an unrealistic
manner. This is because in realistic problems, we may
come across non-precise (fuzzy) hypothesis. For example,
suppose that u is the proportion of a population which
have a disease. We take a random sample of elements
and study the sample for having some idea about u . In
crisp hypothesis testing, one uses the hypotheses of the
form:
0
0.2 H : = u versus
1
: 0. H 2 u = or
0
: 0. H 2 u s
versus
0
: > 0. H 2 u , and so on. However, we would
sometimes like to test more realistic hypotheses. In this
example, more realistic expressions about u would be
considered as: small, very small, large, approxi-
mately 0.2, essentially larger and so on. Therefore,
more realistic formulation of the hypotheses might be
0
: H u is small, versus
1
: H u is not small. We call
such expressions as fuzzy hypotheses.
We define some models, as fuzzy sets of real numbers,
for modeling the extended versions of the simple, the
one-sided, and the two-sided crisp hypotheses to the
fuzzy ones (Akbari and Rezaei, [17]).
Definition 2.1 Let
0
u be a real number and known.
1) Any hypothesis of the form
( )
0
: .3 .3 .3 H cmis cmapproximately cm u u is called to be a
fuzzy simple hypothesis.
2) Any hypothesis of the form
( )
0
: .3 .15 .3 .3 H cmis cmnot cmapproximately cm u u is
called to be a fuzzy two-sided hypothesis.
3) Any hypothesis of the form
( : .3 .15 .5 .15 .15 . H cmis cmessentially cmlarger cmthan cm u
)
0
13cmu is called to be a fuzzy right one-sided hypo-
thesis.
4) Any hypothesis of the form
( : .3 .15 .15 .15 .15 . H cmis cmessentially cmsmaller cmthan cm u
)
0
3cmu is called to be a fuzzy left one-sided hypothesis.
We denote the above definitions by
( ) {
( ) ( )}
0
0 0
.3 : .3 .3
.3 .65 .3 .3
a cm H cmis cmapproximately
cm cm cmis cmH
u
u u u

( ) {
}
0
0
.3 : .3 .3 .
3 .3 .3 1.4
b cm H cmis cmessentially
cmlarger cmthan cm cm
u
u

Copyright 2011 SciRes. OJS
M. G. AKBARI 89

( ) {
}
0
0
.3 : .3 .3 .3
.3 .3 1.6
c cm H cmis cmessentially cmsmaller
cmthan cm cm
u
u

( ) { }
0 0
.3 : .3 .3 .3 .65 d cm H cmis cmapproximately cm cm u u

2.3. Density (Mass) Probability Function

Let X is a RV and let
( ) { }
= : >
X
S x R f x u e 0 be
the support or sample space of X and
( )
( ) ( )
( )
1
0
1
0
d d
d d
H f x
f x
H
o
o
u u
u u
u u u o
u
u u o
e
e
=
} }
} }

where ( ) H u is the membership function of canonical
fuzzy hypothesis and
o
u

is its o -cuts.
We call the new density
(
f x
)
u

as the fuzzy
probability density (mass) function (FPDF) of X
(Akbari and Rezaei [18]). We note that,
( )
0 > f x u

and
( )
( ) ( )
( )
( ) ( )
( )
1
0
1
0
1
0
1
0
d d
d
d d
d d d
d d
1
X X
X
x S x S
x S
H f x
f x x
H
H f x x
H
o
o
o
o
u u
u u
u u
u u
u u u o
u
u u o
u u u o
u u o
e
e e
e
e e
e
=
=
=
} }
} }
} }
} } }
} }

(substitute the summation by integral in discrete cases).
Let ( ) : g x R R be arbitrary function in x . Then
we define
( ) ( ) ( )
( )
( ) ( ) ( )
( )
( ) ( ) ( )
( )
( ) ( ) ( )
( )
1
0
1
0
1
0
1
0
1
0
1
0
d
d d d
d d
d d d
d d
d d
d d
X
X
X
x S
x S
x S
E g X g x f x x
g x H f x x
H
H g x f x x
H
H E g X
H
o
o
o
o
o
o
u
u u
u u
u u
u u
u
u u
u u
u
u u u o
u u o
u u u
u u o
u u o
u u o
e
e e
e
e e
e
e
e
=
=
=
=
}
} } }
} }
} } }
} }
} }
} }

o
)

Let (
1 2
= , , ,
n
X X X X
value
1 2
= ,
be a random sample, with
observed ( ) , ,
n
x x x x , where
i
X has the
FPDF
( )
f x u

with unknown u eO . For testing

( ) :
( )
0 0
1 1
:
0
1
H is H or is
H is H or is
u u u u
u u u u

we state the following definitions:
Definition 2.2 Let ( ) X be a test function. The
probability of type I error of ( ) X is
( )
0
E X
u
o = (

,
and the probability of type II error of ( ) X is
( ) (
1 1
1 1 E X E X
u u
| = = ) ( (



.
Definition 2.3 A teat is said to be a test of level
o if

o o s , where | | 0,1 o e .
we call

o the size of .

3. Sequential Probability Ratio Test

Consider testing a null fuzzy hypothesis against a
alternative fuzzy hypothesis. In other words, suppose a
sample can be drawn from one of two FPDFs and it is
desired to test that the sample came from one distribution
against the possibility that is came from the other. If
denotes the random variables, we want to test
1 2
, , X X
( ) 0
: ~ .
i
H X f
0
u

versus
( ) 0
: ~ .
i 1
X f H u

. Th
o test was of the following form:
e simple
likelihood-rati
0
0
= >
1
0. reject H if k for some constant k
L
s
The sequential test that we propose to consider
em
L
ploys the likelihood-ratios sequentially. Define
( )
( )
( )
( )
( )
( )
( )
( )
( )
0 1 2 0
1 1 1 2
0
1
0
1
1
1
= = =
, , ,
=
m
m m
m
m
i
i
m
i
i
x
L x L x x x
f x
L m
L m
f x

u
u
=
=
=
[
[

for and compute sequentially
, , , L x x x L x
=1, 2, , m
1 2
, , .
dopt the for d
1
k satisfying
0 1
0 < < k k , a
following cedure take observa
fixed
0
k an
pro : tion
1
x and compute
1
; if
1 0
k s , reject
0
H ; if
1 1
k > , accept
0
H ; and if
<
0 1 1
< k k , take obs tion erva
2
x , and comp
2
ute . If
2 0
k s , reject
0
H ; if
2 1
k > , accept
0
H ; and if
<
0 2 1
< k k , take servat , and etc. e idea is
ampling as long as
1
< <
j
k
ob ion x
3
T
0
k
h
to continue s and stop as
soon as
0 j
k s or
1 j
k > , rejecting
0
H if
0 m
k s
and accep
0
ting H if
1
k
m
> . The critic egio
described seque be define as
=1
=
n n
C C

,
where
al r n of the
ntial test can
( ) {
( ) }
0 1 2 1
1 2 0
= : < , , , < ,
=1, 2, , 1, , , ,
n j
n
C x k x x x k
j n x x x

k
Similarly, the acceptance region can be defined as
=1
= .
n n
A A

, where
Copyright 2011 SciRes. OJS
M. G. AKBARI 90


j
k
When we considered the simple likelihood-ratio test
fo
( )
( ) }
1 2 1
1 2 1
, , , < ,
=1, 2,..., 1, , , ,
n
n
A x x x k
j n x x x

>

{ 0
= : < x k
r fixed sample size n , we determined k so that the
test would have preass ned size ig o . We know want to
determine
0
k and
1
k so that the quential probability
ratio test will have preassigned
se
o and | for its
respective sizes of type I and type II e ors. Note that rr
( ) ( )
0 0 0
=1
= =
C
n
n
P reject H H is true L n o

}

and
( ) ( )
0 0 1
=1
= =
A
n
n
P accept H H is false L n |

}
,
where, as before, is a shortened notation for ( )
0
C
n
L n
}

( ) 0
1
d
n
i i
i
C
n
f x x u
=
(

[
} }

.
For fixed o and | , the above equations are two
equations in t two unknown
0
k and
1
k . A solution
of these two equations would give t e sequential
probability ratio test having the desired preassigned error
sizes
he
h
o and | . As might be anticipated, the actual
determ
0
k and
1
k from above equations can
be a major comput onal pr ect.
We note that the sample s
ination of
ize of a sequential
pr
ati oj
obability ratio test is a random variable. The procedure
says to continue sampling until
n
first falls outside the
interval ( )
0 1
, k k . The actual sam e size then depend on
which rved; it is a function of the random
variable
1 2
, , X X and consequently is itself a RV.
Denote it by eally, we would like to know the
distribution of N or at least the expectation of N .
One way of assessing the performance of the sequential
probability ratio test would be to evaluate the expected
sample size that is required under each hypothesis. The
following lemma, given without proof (Lehmann, [20]),
state that the sequential probability ratio test with crisp
hypotheses is an optimal test if performance is measured
using expected sample size. We can similarly prove this
lemma with fuzzy hypotheses based on introduced
FDPF.
Lemma
pl
,
i
x s obse
s
N . Id
3.1 The sequential probability ratio test with
error sizes o and | minimizes both
( )
E N H is e and
0
tru
( )
E N H is true among
followi
1
all tests
which satisfy e ng: th
( )
P H is rejected H is true
0 0
o s ,
( )
0 0
P H is rejected H is true | s , and the expected
sample size is finite.
t the determination of and
th
We noted above tha
0
k
1
k
at defines that particular sequential probability ratio
test which has error sizes o and | is in general
computationally quite difficult. The following lemma
(with simple proof) gives an approximation to
0
k and
1
k .
Lemma 3.2 Let and be defined so that the
sequential probability
0
k
1
k
e ratio t st corresponding to
0
k
and
1
k has error sizes o and | ; then
0
k and
can b pproximated by, sa
0
k' an
1
k' , wh e
1
k
e a y d er
0 1
1
= = .
1
k and k
o o
|

|
' '

Lemma 3.3 Let o' |' and be the error sizes the
sequential probability
of
ratio test defined by
0
k' and
1
k'
given in before lemma. Then o | o | ' ' + s +
Naturally, one would pre se
.
at fer to u h se t quential
probability ratio test having the desired preassigned error
sizes o and | ; however, since it is difficult to to find
the k and corresponding to such a sequential
probability ratio test, instead one can use that sequential
probability ratio test defined by
0 1
k
o' and |' of before
equation and be assured that the the sum f the error
sizes
o
o' and |' is less than or equal to the sum of the
desire rror siz d e es o and | .
The procedure used in e p rforming a se ntial que
probability ratio test is to continue sampling as long as
0 1
< <
m
k k and stop sampling as soon as
0 m
k s or
1 m
<
( )
k > . If
( )
0
1
ln
i
i
z
f x u
i
f x u
(
(
=
(

, an e va
n by the following: continue sam ling as as
ratio test, and let

qui
p
lent test is
long give
i

0 1
=1
ln < < ln
n
i
i
k z k

, and stop sampling as soon as

1
=1
ln
n
i
i
z k > . As before, let N be a
he sa he sequential probability
0
=1
ln
n
i
z k s or

RV denoting t mple size of t

( )
( )
0
1
ln
i
i
i
z
f x u
f x u
(
(
=
(

.
If the sequential probability ratio test leads to rejection

of
0
H , then the RV
0
=1
ln
N
i
i
Z k >

, but
=1
N
i
i
Z

is
close to
0
ln k since
=1
i
i
N
Z

first becam an o e
=1

ere
less th r
h
i ; hence
k , wh
equal to at the observation; hence
0
ln k N t
0
ln
N
E Z k
(

. Sim larly ln
N
i
E Z k
(
~

=
i
i a
~

1
i
( )
0 1
=1
ln 1 ln
N
i
i
E Z k
(
~ +

( )
0
= P H is rejected . Using Wald
,
s equation (Casella
and Berger, [19])
( )
( )
( )
=1 0 1
ln ln
( ) =
N
i
i
E Z
k k
E N
E Z E
(

~

1
i i
Z
+
(
Copyright 2011 SciRes. OJS
M. G. AKBARI 91

we obtain
( )
( )
( )
( )
( )
0 1
0
0
0
ln 1 ln


1
ln 1 ln
1

i
i
k k
E N H is true
E Z N H is true
E Z H is true
o o
o o
o o
| |
+
~
| | | |
+
| |

\ . \ .
~

and
( )
( )
( )
( )
( )
0 1
0
0
0
1 ln ln


1
1 ln ln
1
.

i
i
k k
E N H is flase
E Z H is flase
E Z H is true
| |
o o
| |
| |
+
~
| | | |
+
| |

\ . \ .
~


4. Numerical Examples

In this section, we illustrate the proposed approach for
me distributions and use the ability of package Maple
( Behboodian, [6]) Let
so
6 [21] for this examples.
Exam le 4.1 Taheri and p X
be a continues r.v. with PDF
( ) ( )( ) = 2 2 1 1 0 < <1 0 < <1. f x x x x u u u u +
we want to test
0
1
1
:
3
1
:
2
H is approximatly
H is approximatly
u
u

where the membership functions

0
H and
1
H are de-
fined in the following way:
( )
( )
0
3
1 2
= 2 3 H u u

1
1
3 0 <
<
3 3
0
1
2 0 <
2
1
= 2 2 <1
2
0
otherwise
H
otherwise
u u
u
u u
u u u

We can interpret
0
u

and
1
u

as the value of

1

3
near to and
1

2
near to .
Let = 0.05 o
( )
0
= 2.986 k' , ln ( )
1
ln = 4. ' . Hence, 554 k
( )
= E true
0
N H is 38.556 , and = 39 n , we must take
whereas
( )
= 25.4 H is true
t
1
= 25
Le
32 E N , thus we take
.
Example 4.2
n
( ) , ,
n 1 2
= , X X X X be a random
sample where ( ) ~ ,1
i
X u population, i.e.,
( ) ( )
2 1 1
exp
2
= , f x x x R

e
`
)

2
and ( )
,
i
H
u u u
u s a y hypotheses with me re our fuzz mbership
funct n by:
We can interpret
ions give
( )
0
= 13 12 <13
0
H
otherwise
u u u s

( )
1
11 11 <12
9 9 <10
= 11 10 <11
0
H
otherwise
u u
u u
u u u
s

0
H and
1
H
.
as the value of
and
Let
12 near to 10 to near
. Hence,
( )
0
= 85.155 E N H is true = =0.05 o | ,
and we , whereas must take n = 85
( )
1
= 8 N H is true , thus e take = 88 n .
4.3 Le
7.63
t
E w
Example X be a where a r ndom sample
( )
and = 0.01 | . We obtain
~ ,1
i
X E u pop ulation, i.e.,
( ) ( ) ( )
= > 0, f x x x u u exp , > u u
and ( )
,
i
H u s are our iangular fuzzy parameters with
unctions
tr
membership f
( ) =
i
i
a
a b
u
u

,
i
i i
i i
a a
a a a
H
a a b
b
u
u
u
+
s s

s s +
for .
We can interpret the canonical parameters as having
values that are near to .
Let
, 0 a b >
i
a
,
0
= 8 a ,
1
= 4 a = 0.05 and o = 0.1 | . Hence,
( )
21.325
0
whereas
= E N true H is , and we must take n = 21,
( )
1
= 28.22 E N H is true , thus we take
n . = 28
Exam .4 ple 4 Let X be a R the V from ( ) ~ 0, U u
population, i.e.,
( )
1
= 0 , > 0, f x x u u u
u
and ( )
,
i
H
s s
u s are our trapezoidal fuzzy parameters with
mem unctions given by: bership f
Copyright 2011 SciRes. OJS
M. G. AKBARI

Copyright 2011 SciRes. OJS
92
( )
( )
( )
2
2
= 1
i i i
a b
x x
2 2
2
.
2
i
i i
i
i i
x a
a
x a x
a
H
x b
b
x X b
b
u
u
u
u
u
+
s s

+
+ s s +

for
Let and . If
u

s s +

[6] S. M. Taheri and J. Behboodian, Neyman-Pearson Lem-
ma for Fuzzy Hypotheses Testing, Metrika, Vol. 49, No.
1, 1999, pp. 3-17. doi:10.1007/s001840050021

[7] H. Torabi, J. Behboodian and S. M. Taheri, Neyman-

Pearson Lemma for Fuzzy Hypotheses Testing with-
Vague Data, Metrika, Vol. 64, No. 3, 2006, pp. 289-304.
doi:10.1007/s00184-006-0049-8
[8] P. Filzmoser and R. Viertl, Testing Hypotheses with
Fuzzy Data: The Fuzzy p-value, Metrika, Vol. 59, No. 1,
2004, pp. 21-29. doi:10.1007/s001
, 0 a b > .
840300269
= =1 a b ,
0
= 5 x
1
= 3 x = = 0.05 o | ,

then
( )
= 8 true
0
.75 E N H is
eas
[9] R. Viertl, Univariate Statistical Analysis with Fuzzy
Data, Computational Statistics and Data Analysis, Vol.
51, No. 1, 2006, pp. 133-147.
,
wher
and we must take = 9 n ,
( )
1
is true = 9. E N H 11, thus we ta

onclu ons

paper, an new appr est o
potheses based on fuz for one
mple and two-sample when the available data are crisp,
r this paper, it sound the introdu
ethod is very simple and applicable in the statistics and
ypotheses based on intuitionistic fuzzy
hy
ce Ltd, New York, 1995, pp. 307-316.
An Approach to Fuzzy Hypothesis Test-
ke = 9 n .
doi:10.1016/j.csda.2006.04.002
[10] J. J. Buckley, Fuzzy Probabilities: New Approach and
Applications, Springer-Verlag
[11] J. J. Buckley, Fuzzy Probability and
5. C si
, Berlin, 2005.
Statistics, Springer
-values in
.1093/biomet/asm001
In this oach for sequential t f
fuzzy hy zy hypotheses -
-Verlag, Berlin, 2006.
[12] E. A. Thompson and C. J. Geyer, Fuzzy p
sa
is presented. As fo ced
Latent Variable Problems, Biometrika, Vol. 94, No. 1,
2007, pp. 49-60. doi:10
m
other sciences.
Extension of the proposed method to test the variance,
correlation and parameters of linear models (regression
models), design of experiment is a potential area for the
future work. Furthermore, we can construct sequential
test of fuzzy h

-0339-3
[13] S. M. Taheri and M. Arefi, Testing Fuzzy Hypotheses
Based on Fuzzy Statistics, Soft Computing, Vol. 13, No.
6, 2009, pp. 617-625. doi:10.1007/s00500-008
[14] A. Parchami, S. M. Taheri and M. Mashinchi, Fuzzy
p-value in Testing Fuzzy Hypotheses with Crisp Data,
Statistical Papers, Vol. 51, No. 1, 2010, pp. 209-226.
doi:10.1007/s00362-008-0133-4
[15] P. Billingsley, Probability and Measure, 2nd Edition,
John and Wiley, New York, 1995.
[16] G. Klir and B. Yuan, Fuzzy Sets a
potheses or fuzzy data for the parameters of interest.

6. References

[1] M. Delgado, J. L. Verdegay and M. A. Vila, Testing
Fuzzy-Hypotheses: A Bayesian Approach, In: M. M.
Gupta, Ed., Approximate Reasoning in Expert System
nd Fuzzy Logic-Theory
s with Applications, Vol. 37, No. 8, 2010, pp.
and Applications, Prentice-Hall, Upper Saddle River,
1995.
s,
[17] M. G. Akbari and A. Rezaei, Bootstrap Testing Fuzzy
Hypotheses and Observations on Fuzzy Statistic, Expert
System
Elsevier Scien
2] B. F. Arnold, [
ing, Metrika, Vol. 44, No. 1, 1996, pp. 119-126.
doi:10.1007/BF02614060
[3] B. F. Arnold, Testing Fuzzy Hypothesis with Crisp Da-
ta, Fuzzy Sets and Systems, Vol. 94, No. 3, 1998, pp.
323-333. doi:10.1016/S0165-0114(96)00258-8
5782- 5787. doi:10.1016/j.eswa.2010.02.030
[18] M. G. Akbari and A. Rezaei, Discrete and Continuous
Random Variables with Fuzzy Parameter, Far East
Journal of Theoretical Statistics, Online, 2009.
Chapman
Canada.
[19] G. Casella and R. L. Berger, Statistical Inference, 2nd
Edition, Duxbury Press, Belmont, 2002.
[20] E. L. Lehmann, Testing Statitical Hypotheses,
[4] M. Holena, Fuzzy Hypotheses for GUHA Implications,
Fuzzy Sets and Systems, Vol. 98, No. 1, 1998, pp. 101-125.
doi:10.1016/S0165-0114(96)00369-7
[5] M. Holena, Fuzzy Hypotheses Testing in the Framework
of Fuzzy Logic, Fuzzy Sets and Systems, Vol. 145, No. 2,
2004, pp. 229-252.
and Hall, London, 1994.
[21] Maple 6, Waterloo Maple Inc. Waterloo,
002 doi:10.1016/S0165-0114(03) 08-2