Advances and Applications in Statistics

© 2024 Pushpa Publishing House, Prayagraj, India

https://pphmjopenaccess.com
http://dx.doi.org/10.17654/0972361724023
Volume 91, Number 4, 2024, Pages 439-450 P-ISSN: 0972-3617

VAGUE DATA ANALYSIS USING SEQUENTIAL TEST

Abstract

Objective. The existing sequential test using Bernoulli distribution

can only be applied when no uncertainty/indeterminacy is found in
testing the hypothesis. This paper introduces neutrosophic Bernoulli
distribution and sequential test using the distribution.

Method. The operational procedure of the proposed test will be

introduced and applied for testing the hypothesis in the presence of
uncertainty.
Received: October 29, 2023; Accepted: December 16, 2023
2020 Mathematics Subject Classification: 97K80.
Keywords and phrases: sequential test, neutrosophy, classical test, industry, Bernoulli
distribution.
How to cite this article: Muhammad Aslam and Muhammad Saleem, Vague data analysis
using sequential test, Advances and Applications in Statistics 91(4) (2024), 439-450.
http://dx.doi.org/10.17654/0972361724023
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Published Online: February 12, 2024
440 Muhammad Aslam and Muhammad Saleem
Results. The advantages of the proposed test will be discussed using
manufacturing data. From the comparison and simulation studies, it is
found that the proposed test is efficient than the existing test under
classical statistics.

Conclusions. The proposed test may be more economic and time-

saving as the decision may make on the basis of the first sample.
Therefore, the proposed test is efficient, economical and adequate than
the existing test.

1. Introduction

Usually, the statistical tests give decisions about the acceptance or

rejection of the null hypothesis using the information obtained from a single
sample. In practice, the decision-makers sometimes are unable to reach the
decision using single sample of information. In case, when decision-makers
are in-decision on the basis of the first sample, a second sample is taken and
the process is continued until a decision is reached. The sequential test using
the Bernoulli distribution is also applied when no decision about the null
hypothesis is made on the first sample. Causey [13] provided the exact
calculation of the test. This test is applied for testing the parameter of the
Bernoulli distribution. Kanji [26] discussed the sequential test for Bernoulli
distribution. Bartroff and Song [10] studied the power of the sequential tests.
Pereira et al. [31] proposed the sequential test using the Bayesian approach.
Pramanik et al. [32] presented the modification of the sequential test. More
details can be read in Gardonyi and Samu [19], and Pramanik et al. [32]
presented the modified forms of SPRT. More applications of SPRT can be
seen in Wijsman [44], Thomas [41], Bacanli and Icen [9], and Pan et al. [28].
The fuzzy-based statistical tests attract the decision-makers when
uncertainty/indeterminacy is found in the observations or parameters. The
fuzzy-based tests are found to be effective when testing of hypothesis is done
in uncertainty. Kacprzyk et al. [24] and Taheri and Hesamian [39] pointed
out the fields where imprecise observations can be recorded. Torabi and
Behboodian [42], Torabi and Mirhosseini [43], Parchami [29], Talukdar and
Baruah [40], İçen et al. [23], Kahraman et al. [25], Dubois and Prade [18],
 Vague Data Analysis Using Sequential Test 441
Grzegorzewski [20], Denoeux et al. [17], Grzegorzewski [21], Taheri and
Arefi [38], Taheri and Hesamian [37], Kacprzyk et al. [24], Taheri and
Hesamian [39], Shafiq et al. [33], Grzegorzewski and Śpiewak [22] and,
Chukhrova and Johannssen [16] applied statistical tests on imprecise data.
Smarandache [35] introduced neutrosophic logic and showed the
efficiency over the fuzzy-based logic and interval-based analysis. The
neutrosophic logic gives information about the measure of indeterminacy in
addition to the fuzzy logic. Broumi and Smarandache [12], Abdel-Basset et
al. [1], Peng and Dai [30], Shahin et al. [34], Broumi et al. [11], Abdel-
Basset et al. [2, 3] and Nabeeh et al. [27] showed the efficiency of
the neutrosophic logic over the existing logic. Smarandache [36] used
neutrosophic logic to develop neutrosophic statistics to analyze imprecise
data. Chen et al. [14, 15], and Aslam [4-7] showed that neutrosophic
statistics is efficient than classical statistics.

The sequential test using Bernoulli distribution can be applied only

when the parameter of the distribution is certain, precise and exact. The
existing sequential test using Bernoulli distribution under classical statistics
cannot be applied when the observation/parameter is indeterminate. By
exploring the literature, no work is found on sequential test using Bernoulli
distribution under neutrosophic statistics. In this paper, we introduce
neutrosophic Bernoulli distribution first. We present the design of a
sequential test using Bernoulli distribution under neutrosophic statistics. The
application of the proposed test will be given in the manufacturing industry.
From the study, it is expected that the proposed will be adequate than the
existing sequential test using the Bernoulli distribution.

2. Method

Suppose that X 1N , X 2 N , X 3 N , ..., X nN is independently and identically

neutrosophic random variable which follows the neutrosophic Bernoulli
distribution with the neutrosophic probability density function (npdf), say
x
f0 NN is given by
442 Muhammad Aslam and Muhammad Saleem

I N f 0 ∈ [I L f 0 , IU f 0 ].
x x x
f 0 NN = f 0 N + f 0 N I N f 0 ; (1)

x
Note that f0 N represents the Bernoulli distribution under classical

statistics, f 0 N I N f 0 presents the indeterminate part and I N f 0 ∈ [I L f 0 , IU f 0 ]

x

is the measure of indeterminacy. The current neutrosophic form reduces

to Bernoulli distribution (BD) under classical statistics when I L f 0 = 0,
see Aslam [5]. Using the neutrosophic theory information, the npdf of
neutrosophic Bernoulli distribution (NBD) is given by

f 0 NN = θ NN (1 − θ NN )1− x N + θ NN (1 − θ NN )1− xN I N f 0 ;
x x x x x

I N f 0 ∈ [I L f 0 , IU f 0 ], (2)

where [0, 0] ≤ θ NN ≤ [1, 1]. The proposed NBD reduces to BD when

x

I L f 0 = 0. The proposed NBD can be expressed as follows:

f 0 NN = θ NN (1 − θ NN )1− x N (1 + I N f 0 ); I Nf 0 ∈ [I L f 0 , IU f 0 ].
x x x
(3)

Suppose that the decision-makers are interested to apply the proposed

test when uncertain about the specified value of the unknown parameter.
Let θ0 N = θ0 N + θ0 N I θ0 N and I θ0 N ∈ [I θ0 L , I θ0U ] be the neutrosophic

form against the null hypothesis H 0 of the unknown parameter, where

θ0 N is the exact value, θ N I θ0 N is the indeterminate part and

I θ0 N ∈ [I θ0 L , I θ0U ] is an indeterminate interval. Let θ1N = θ1N + θ1N I θ1N

and I θ1N ∈ [I θ1L , I θ1U ] be the neutrosophic form against the alternative

hypothesis H1 of the unknown parameter, where θ1N is the exact

value, θ1N I θ0 N is the indeterminate part and I θ1N ∈ [I θ1L , I θ1U ] is an

i =1 X iN
mN
indeterminate interval. Let SmN = be the sum of neutrosophic

random variables, α N be a type-I and β N be a type-II error. The

 Vague Data Analysis Using Sequential Test 443
neutrosophic lower limit, say amN ∈ [amL , amU ] of the proposed test is
given by

βN 
log 
amN =  1 − αN 
 θ1N (1 + I θ1N )   1 − {θ1N (1 + I θ1N )} 
log  − log 
 θ0 N (1 + I θ0 N )   1 − {θ0 N (1 + I θ0 N )} 

 1 − {θ0 N (1 + I θ0 N )} 
m N log 
 1 − {θ1N (1 + I θ1N )} 
+ . (4)
 θ1N (1 + I θ1N )   1 − {θ1N (1 + I θ1N )} 
log  − log 
 θ0 N (1 + I θ0 N )   1 − {θ0 N (1 + I θ0 N )} 

The neutrosophic upper limit, say rmN ∈ [rmL , rmU ] of the proposed test
is given by
1 − βN 
log 
rmN =  αN 
 θ1N (1 + I θ1N )   1 − {θ1N (1 + I θ1N )} 
log  − log 
 θ0 N (1 + I θ0 N )   1 − {θ0 N (1 + I θ0 N )} 

 1 − {θ0 N (1 + I θ0 N )} 
m N log 
 1 − {θ1N (1 + I θ1N )} 
+ . (5)
 θ1N (1 + I θ1N )   1 − {θ1N (1 + I θ1N )} 
log  − log 
 θ0 N (1 + I θ0 N )   1 − {θ0 N (1 + I θ0 N )} 

The proposed limits are the generalization of the limits under classical
statistics. The proposed two limits under neutrosophic statistics reduce to
limits under classical statistics when I θ0 N = 0 and I θ1N = 0.

The testing process of the proposed test is stated as follows:

Step 1. State H 0 : θ0 N = θ0 N (1 + I θ0 N ); I θ0 N ∈ [I θ0 L , I θ0U ] vs. the

alternative hypothesis θ1N = θ1N (1 + I θ1N ); I θ1N ∈ [I θ1L , I θ1U ].

Step 2. Specify α N and β N and compute the values of amN ∈

[amL , amU ] and rmN ∈ [rmL , rmU ].
444 Muhammad Aslam and Muhammad Saleem
Step 3. The decision-makers failed to reject H 0 if SmN < amN and
reject H 0 if SmN ≥ rmN . The decision-makers should continue to take
samples if amN < SmN < rmN .

3. Application

The application of the proposed test will be given using the

information obtained from the manufacturing industry. Suppose that θ N is
the proportion of defective items during the manufacturing process, 1 − θ N
is the non-defective product and the manufacturer is uncertain about the
defective and non-defective items with I θ N . Based on the information, the
manufacturer may willing to adopt the new manufacturing process if θ N ≤
θ0 N (1 + I θ0 N ); I θ0 N ∈ [I θ0 L , I θ0U ] and reject θ N ≥ θ1N (1 + I θ1N );
I θ1N ∈ [I θ1L , I θ1U ] and he may in-decision when θ0 N (1 + I θ0 N ); I θ0 N ∈
[I θ0 L , I θ0U ] < θ N < θ1N (1 + I θ1N ); I θ1N ∈ [I θ1L , I θ1U ]. Suppose that
θ0 N = 0.05, θ1N = 0.30, I θ0U = 0.05, I θ1N = 0.30, α N = 0.05 and
β N = 0.10. For the given data, amN ∈ [amL , amU ] and rmN ∈ [rmL , rmU ]
are calculated by
amN = −0.9205 + 0.1800 mN ,

rmN = 1.1818 + 0.1800 mN .

The testing process of the proposed test is stated as follows:

Step 1. State H 0 : θ0 N = 0.0525 vs. the alternative hypothesis

θ1N = 0.39.

Step 2. Specify α N = 0.05 and β N = 0.10 and compute the values of

amN = −0.9205 + 0.1800mN and rmN = 1.1818 + 0.1800mN .

Step 3. The decision-makers failed to reject H 0 if SmN < −0.9205 +

0.1800mN and reject H 0 if SmN ≥ 1.1818 + 0.1800m N . The decision-
 Vague Data Analysis Using Sequential Test 445
makers should continue to take samples if − 0.9205 + 0.1800mN < S mN
< 1.1818 + 0.1800mN .

The operational process of the proposed test for the real example is
shown in Figure 1.

Figure 1. The operational process for the example.

4. Simulation

In this section, a simulation study is performed to check the effect of

the measure of uncertainty/indeterminacy on the lower and upper limits
of the proposed test. As mentioned earlier, two measures of uncertainty
I θ0 N ∈ [I θ0 L , I θ0U ] and I θ1N ∈ [I θ1L , I θ1U ] are associated with null
hypothesis and alternative hypothesis. In the simulation study, various values
of I θ0 N and I θ1N are considered. The values of amN and rmN are reported

in Table 1. From Table 1, it can be seen that the intercept of amN decreases
as the values of I θ0 N and I θ1N increase. On the other hand, the values of

rate of change (slope) increase as the values of I θ0 N and I θ1N increase. We

also note the decreasing trends in intercept of rmN as the values of I θ0 N

and I θ1N increase. The increasing trend is also noted in the slope of rmN as
446 Muhammad Aslam and Muhammad Saleem
the values of I θ0 N and I θ1N increase. From the simulation study, it can be

concluded that as I θ0 N and I θ1N increase, the two limits become narrow
and the change of continuing sampling is reduced.

Table 1. Effect of measure of uncertainty on limits

I θ0 N I θ1N amN rmN

0 0 −1.0735 + 0.1456 m N 1.3782 + 0.1456 m N

0.01 0.02 −1.0644 + 0.1482 m N 1.3665 + 0.1482 m N

0.02 0.03 −1.0625 + 0.1497 m N 1.3641 + 0.1497 m N

0.03 0.04 −1.0606 + 0.1512 m N 1.3617 + 0.1512 m N

0.04 0.05 −1.0588 + 0.1528 m N 1.3593 + 0.1528 m N

0.05 0.06 −1.0569 + 0.1543 m N 1.3569 + 0.1543 m N

0.06 0.07 −1.0550 + 0.1559 m N 1.3545 + 0.1559 m N

0.07 0.08 −1.0512 + 0.1590 m N 1.3497 + 0.1590 m N

0.09 0.10 −1.0493 + 0.1605 m N 1.3472 + 0.1605 m N

5. Comparative Study

The proposed sequential test for the Bernoulli distribution is the

extension of the existing test under classical statistics. The proposed test
reduces to the classical test when I θ0 N = 0 and I θ1N = 0. In this section,
the comparison of the proposed test is given with the existing test in terms of
a measure of uncertainty. The limits of the classical test when I θ0 N = 0 and

I θ1N = 0 are shown in Table 1. The limits for various values of I θ0 N and

I θ1N are also shown in Table 1. From Table 1, it can be seen that the gap

between two limits reduces as the values of I θ0 N and I θ1N increase. On the

other hand, a big gap can be noted in limits when I θ0 N = 0 and I θ1N = 0.
From the study, it can be seen that as the gap between two limits decreases,
it also decreases the chance of continued sampling. In addition, the
 Vague Data Analysis Using Sequential Test 447
comparison shows that there is a high chance of resampling in the case of the
existing test. In nutshell, the proposed test may be more economic and time-
saving as the decision may make on the basis of the first sample. Therefore,
the proposed test is efficient, economical and adequate than the existing test.

6. Concluding Remarks

We introduced neutrosophic Bernoulli distribution and used it in

designing the sequential test. The proposed test was found to be in more
generalized form than the existing sequential test. The operational procedure
of the proposed test was explained with the help of a diagram. The
application of the proposed test was given using industrial data. From
the comparison and simulation studies, the proposed test was found more
efficient than the existing sequential test. The proposed test has the
limitations that it can be applied only when the data is obtained from
neutrosophic Bernoulli distribution. The proposed test using a cost model
can be studied in future.

References

[1] M. Abdel-Basset, A. Atef and F. Smarandache, A hybrid neutrosophic multiple

criteria group decision making approach for project selection, Cognitive Systems
Research 57 (2019), 216-227.
[2] M. Abdel-Basset, M. Mohamed, M. Elhoseny, F. Chiclana and A. E.-N. H. Zaied,
Cosine similarity measures of bipolar neutrosophic set for diagnosis of bipolar
disorder diseases, Artificial Intelligence in Medicine 101 (2019), 101735.

[3] M. Abdel-Basset, N. A. Nabeeh, H. A. El-Ghareeb and A. Aboelfetouh, Utilising

neutrosophic theory to solve transition difficulties of IoT-based enterprises,
Enterprise Information Systems 14 (2020), 1304-1324.

[4] M. Aslam, Introducing Kolmogorov-Smirnov Tests under Uncertainty: An

Application to Radioactive Data, ACS Omega, 2019a.

[5] M. Aslam, Neutrosophic analysis of variance: application to university students,

Complex and Intelligent Systems 5(4) (2019b), 403-407.
448 Muhammad Aslam and Muhammad Saleem
[6] M. Aslam, Design of the Bartlett and Hartley tests for homogeneity of variances
under indeterminacy environment, Journal of Taibah University for Science
14(1) (2020), 6-10.
[7] M. Aslam, Radar data analysis in the presence of uncertainty, European Journal of
Remote Sensing 54(1) (2021a), 140-144.
[8] M. Aslam, Testing average wind speed using sampling plan for Weibull
distribution under indeterminacy, Scientific Reports 11(1) (2021b), 1-9.
[9] S. Bacanli and D. Icen, Sequential probability ratio test of correlation coefficient
using fuzzy hypothesis testing, Open Journal of Statistics 3 (2013), 195-199.
[10] J. Bartroff and J. Song, Sequential tests of multiple hypotheses controlling type I
and II familywise error rates, J. Statist. Plann. Inference 153 (2014), 100-114.
[11] S. Broumi, M. Talea, A. Bakali, F. Smarandache and K. Ullah, Bipolar
neutrosophic minimum spanning tree, Smart Application and Data Analysis for
Smart Cities (SADASC’18), 2018. Available at
SSRN: https://ssrn.com/abstract=3127519.
[12] S. Broumi and F. Smarandache, Correlation coefficient of interval neutrosophic
set, Applied Mechanics and Materials 436 (2013), 511-517.
[13] B. D. Causey, Exact calculations for sequential tests based on Bernoulli trials,
Comm. Statist. Simulation Comput. 14(2) (1985), 491-495.
[14] J. Chen, J. Ye and S. Du, Scale effect and anisotropy analyzed for neutrosophic
numbers of rock joint roughness coefficient based on neutrosophic statistics,
Symmetry 9(10) (2017), 208.
[15] J. Chen, J. Ye, S. Du and R. Yong, Expressions of rock joint roughness
coefficient using neutrosophic interval statistical numbers, Symmetry 9(7) (2017),
123.
[16] N. Chukhrova and A. Johannssen, Nonparametric fuzzy hypothesis testing for
quantiles applied to clinical characteristics of COVID‐19, International Journal of
Intelligent Systems 36(6) (2021), 2922-2963.
[17] T. Denoeux, M.-H. Masson and P.-A. Hébert, Nonparametric rank-based statistics
and significance tests for fuzzy data, Fuzzy Sets and Systems 153(1) (2005), 1-28.
[18] D. Dubois and H. Prade, Ranking fuzzy numbers in the setting of possibility
theory, Inform. Sci. 30(3) (1983), 183-224.
[19] G. Gardonyi and K. Samu, An enhanced evaluation method of sequential
probability ratio test, Math. Probl. Eng. Volume 2019, Article ID 4724507.
https://doi.org/10.1155/2019/4724507.
 Vague Data Analysis Using Sequential Test 449
[20] P. Grzegorzewski, Statistical inference about the median from vague data, Control
Cybernet. 27 (1998), 447-464.
[21] P. Grzegorzewski, k‐sample median test for vague data, International Journal of
Intelligent Systems 24(5) (2009), 529-539.
[22] P. Grzegorzewski and M. Śpiewak, The sign test and the signed‐rank test for
interval‐valued data, International Journal of Intelligent Systems 34(9) (2019),
2122-2150.
[23] D. İçen, S. Bacanlı and S. Günay, Fuzzy approach for group sequential
test, Advances in Fuzzy Systems Volume 2014, Article ID 896150.
https://doi.org/10.1155/2014/896150.
[24] J. Kacprzyk, E. Szmidt, S. Zadrożny, K. T. Atanassov and M. Krawczak,
Advances in fuzzy logic and technology 2017, Proceedings of: EUSFLAT-2017-
The 10th Conference of the European Society for Fuzzy Logic and Technology,
September 11-15, 2017, Warsaw, Poland IWIFSGN’2017-The Sixteenth
International Workshop on Intuitionistic Fuzzy Sets and Generalized Nets,
September 13-15, 2017, Warsaw, Poland, Springer, Vol. 2, 2017.
[25] C. Kahraman, C. E. Bozdag, D. Ruan and A. F. Özok, Fuzzy sets approaches to
statistical parametric and nonparametric tests, International Journal of Intelligent
Systems 19(11) (2004), 1069-1087.
[26] G. K. Kanji, 100 Statistical Tests, Sage Publishing, 2006.
[27] N. A. Nabeeh, F. Smarandache, M. Abdel-Basset, H. A. El-Ghareeb and A.
Aboelfetouh, An integrated neutrosophic-TOPSIS approach and its application to
personnel selection: a new trend in brain processing and analysis, IEEE Access
7 (2019), 29734-29744.
[28] J. Pan, Y. Li and V. Y. Tan, Asymptotics of sequential composite hypothesis
testing under probabilistic constraints, 2021. arXiv preprint arXiv:2106.00896.
[29] A. Parchami, Fuzzy decision in testing hypotheses by fuzzy data: two case studies,
Iran. J. Fuzzy Syst. 17(5) (2020), 127-136.
[30] X. Peng and J. Dai, Approaches to single-valued neutrosophic MADM based
on MABAC, TOPSIS and new similarity measure with score function, Neural
Computing and Applications 29(10) (2018), 939-954.
[31] C. A. D. B. Pereira, E. Y. Nakano, V. Fossaluza, L. G. Esteves, M. A. Gannon and
A. Polpo, Hypothesis tests for Bernoulli experiments: ordering the sample space
by Bayes factors and using adaptive significance levels for decisions, Entropy
19(12) (2017), 696.
450 Muhammad Aslam and Muhammad Saleem
[32] S. Pramanik, V. E. Johnson and A. Bhattacharya, A modified sequential
probability ratio test, J. Math. Psych. 101 (2021), 102505.
[33] M. Shafiq, M. Atif and R. Viertl, Generalized likelihood ratio test and Cox’s
F-test based on fuzzy lifetime data, International Journal of Intelligent Systems
32(1) (2017), 3-16.
[34] A. Shahin, Y. Guo, K. Amin and A. A. Sharawi, A novel white blood cells
segmentation algorithm based on adaptive neutrosophic similarity score, Health
Information Science and Systems 6(1) (2018), 1.
[35] F. Smarandache, Neutrosophy Neutrosophic Probability, Set, and Logic,
ProQuest Information and Learning, Ann. Arbor, Michigan, USA, Vol. 105, 1998,
pp. 118-123.
[36] F. Smarandache, Introduction to Neutrosophic Statistics, Sitech and Education
Publisher, Romania-Educational Publisher, Craiova, Columbus, Ohio, USA,
2014.
[37] S. Taheri and G. Hesamian, A generalization of the Wilcoxon signed-rank test and
its applications, Statist. Papers 54(2) (2013), 457-470.
[38] S. M. Taheri and M. Arefi, Testing fuzzy hypotheses based on fuzzy test statistic,
Soft Computing 13(6) (2009), 617-625.
[39] S. M. Taheri and G. Hesamian, Non-parametric statistical tests for fuzzy
observations: fuzzy test statistic approach, International Journal of Fuzzy Logic
and Intelligent Systems 17(3) (2017), 145-153.
[40] R. Talukdar and H. K. Baruah, A two sample sequential t-test with fuzzy
observations, Appl. Math. Sci. 4(68) (2010), 3361-3374.
[41] E. A. Thomas, A note on the sequential probability ratio test, Psychometrika
40(1) (1975), 107-111.
[42] H. Torabi and J. Behboodian, Sequential probability ratio test for fuzzy
hypotheses testing with vague data, Austrian Journal of Statistics 34(1) (2005),
25-38.
[43] H. Torabi and S. Mirhosseini, Sequential probability ratio tests for fuzzy
hypotheses testing, Appl. Math. Sci. 3(33) (2009), 1609-1618.
[44] R. A. Wijsman, Examples of exponentially bounded stopping time of invariant
sequential probability ratio tests when the model may be false, Vol. 1: Theory of
Statistics, University of California Press, 1972, pp. 109-128.