Available online at www.sciencedirect.

com

ScienceDirect
Procedia Computer Science 102 (2016) 385 – 389

12th International Conference on Application of Fuzzy Systems and Soft Computing, ICAFS
2016, 29-30 August 2016, Vienna, Austria

Z-Regression analysis in psychological and educational researches

Fahreddin Sadikoglua*, Oleg Huseynovb, Konul Memmedovac*
a
Department of Electrical and Electronic Engineering, Near East University, Nicosia, North. Cyprus,
b
Department of Computer Engineering, Azerbaijan State Oil and Industry University, Baku, Azerbaijan
c
Department of Psychological Counseling and Guidance, Near East University, Nicosia, North. Cyprus

Abstract

Educational achievement of students is closely related to psychological determinants. Information on relation between
psychological parameters such as motivation, attention, anxiety and educational achievement of students is naturally
characterized by fuzziness and partial reliability. In order to obtain information, we use experiments conducted by using Pilates
exercises. The obtained information is formalized by using Z-number concept which allows describing fuzzy and partially
reliable information. Thus, we use Z-number valued regression model to describe dependence of educational achievement on
motivation, attention, and anxiety. An example is used to show validity of the model.
2016The
© 2016 TheAuthors.
Authors.Published
Publishedbyby Elsevier
Elsevier B.V.B.V.
This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Organizing Committee of ICAFS 2016.
Peer-review under responsibility of the Organizing Committee of ICAFS 2016
Keywords: Psychological parameters; motivation; attention; anxiety; Z-number; Z-regression

1. Introduction

Joseph Pilates, the founder of Pilates exercises, uses mental focus and specific breathing of yoga in line with the
physicality of gymnastics and other sports1-4. The basic Pilates principles are centering, concentration, control,
precision, flow, and breath 5.
In this paper we consider regression-analysis based modeling of impact of Pilates exercises on psychological
determinants as motivation, attention, anxiety affecting educational achievement. Regression analysis problems are
among the most used modeling problems in engineering and social sciences. Constructing adequate mathematical
models requires accounting for the fact that real-world problems are characterized by uncertainty. The reasons are

* Corresponding author.
E-mail address: fahreddin.sadikoglu@neu.edu.tr

1877-0509 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the Organizing Committee of ICAFS 2016
doi:10.1016/j.procs.2016.09.416
386 Fahreddin Sadikoglu et al. / Procedia Computer Science 102 (2016) 385 – 389

complexity of real processes, incomplete relevant information, vagueness of future etc. The attempts taken to
account for real-world uncertainty while constructing mathematical models has a long history. Nowadays a wide
spectrum of regression models with numeric, interval, and fuzzy information exist, e.g6,7,8,9,10 .
The most adequate existing framework to account for real world uncertainty is a combination of possibilistic and
probabilistic information. Zadeh termed such information as a bimodal information11. In order to formalize bimodal
information, Prof. Zadeh suggested a concept of a Z-number12. A Z-number is an ordered pair of fuzzy numbers
Z ( A, B ) used to describe a value of a variable X , where A is a fuzzy constraint on a value of X and B is a
fuzzy constraint on a value of probability measure of A . The key issue is that B plays a role of a reliability of A .
Application of Z-numbers in the fields of decision making, control, reasoning, modeling and other fields is
considered in 13,14,15 . However, in those works dealing with Z-numbers is reduced to computation with fuzzy
numbers that leads to loss of information.
In 16 authors suggested general and computationally effective approach to computations with discrete Z-numbers
as an alternative to the statistical approach for modelling psychological process. The proposed approach allows
dealing with Z-numbers directly.
Let us now explain why information relevant to the considered behavioral determinants and educational
achievement fall within the framework of Z-numbers18.
Motivation. Motivation is defined as the process that initiates, guides, and maintains goal-oriented behaviors. It
involves the biological, emotional, social, and cognitive forces that activate behavior. Naturally, a level of
motivation is described linguistically as a perceived and not measured value. From the other side, the used linguistic
evaluation, are actually partially reliable due to restricted competence. Indeed, motivation is a complex phenomena
inspired by various factors, and fully reliable evaluation can hardly be done.
Anxiety. Anxiety is one of the major psychological variables which is considered as an important part of
personality development. Anxiety is a physiological state characterized by somatic, emotional, cognitive, and
behavioral components4. As one can see, a level of anxiety can also be described by using linguistic evaluation. On
the other hand, any linguistic evaluation is subject to our restricted competence. We cannot uncover actual
psychological condition of a human being, and especially its future dynamics

2. The Z-valued regression model

The set of the observed data expressed as Z-numbers are given describing relation between the independent
behavioral determinants and the dependent educational achievement (Table 1)17.

Table 1. Z-number representation of psychological parameters.

Data Inputs Output

Motivation Attention Anxiety Achievement
(L,U) (L,U) (H,U) (L,U)
(L,U) (M,U) (M,U) (M,U)
... ...
20 (H,U) (L,U) (L,U) (L,U)

Here uppercase letters describe linguistic terms: H-High; L-Low; M- Medium; G-Good; E-Excellence; U-
Usually; P-Plausible; R-Rare. The fuzzy sets related to these linguistic labels are given in
The problem is to construct regression model for fitting the observed Z-valued data. The following Z-valued
linear regression model is considered:

Z y M ( Z x1 , Z x2 , Z x3 ) a1 Z x1  a2 Z x2  a3 Z x3 (1)
 Fahreddin Sadikoglu et al. / Procedia Computer Science 102 (2016) 385 – 389 387

Here Z x1 , Z x2 , Z x3 describe the Z-valued independent variables of the motivation, anxiety and attention; Z y M is
the Z-valued dependent variable describing educational achievement, are numeric coefficients of the model
.
ai , i 1, 2,3 .
A problem of construction of model (1) consists in determination of coefficients ai , i 1, 2,3 given observed
Z-valued input data Z xik and Z-valued output data Z yk , k 1,..., K .
As a criterion of adequacy of regression model (1), we suggest to use the following similarity measure to
compute closeness of Z-valued output of model (1) and Z-valued observed data:

1 K
S ( Z yM , Z y ) ¦ Jk ,
Kk1
(2)

where J k J ( Z yM, k , Z y , k ) Jaccard index18.

Obviously, one has J ( Z yM, k , Z y , k )  [0,1] . According to this measure adopted from20 each k-th Z-valued model (1)
output Z yM, k ( AyM, k , ByM, k ) and k-th Z-valued observed data Z y , k ( Ay , k , B y , k ) are compared component wise as two-
dimensional fuzzy vectors . The overall similarity (2) is obtained as the average one over all the K pairs of Z-valued
data Z yM, k ( AyM, k , ByM, k ) and Z y , k ( Ay , k , B y , k ) , k 1,..., K . In the next section we provide a solution approach for
solving the considered problem of construction of (1).

3. Statement of problem and solution method

The problem of construction of the consider Z-valued regression model can be stated as follows.

1 K
¦ J k o max
Kk1
(3)

Z y M ( Z x1 , Z x2 , Z x3 ) a1 Z x1  a2 Z x2  a3 Z x3 (4)

Z yM, k ( AyM, k , ByM, k ), Z xi , k ( AxMi , k , BxMi , k ), i 1,..., N , k 1,..., K

The procedures of the suggested solution approach are given below.

First step. At the first step randomly choose ai , i 1,..., N .
Second step. At the second step one should compute the Z-output of the regression model (1).
Third step. At the third step we compute the Jaccard index based similarity measure (2) and then the overall
criterion (3). For this purpose one needs to use operations over Z-numbers such as a scalar multiplication of Z-
number, square of a Z-number, addition of Z-numbers and multiplication of Z-numbers. A general framework for
computations over Z-numbers is suggested in Refs. 12,20,21.
Fourth step. At this step we need to use the Newton method based optimization algorithm to maximize overall
criterion (3).

4. Results of simulation and Discussion

In accordance with the approach suggested in Section III, let us construct the Z-valued regression model (1).
At the second step one should compute the Z-output of the regression model (1).
At the third step we compute the Jaccard index based similarity measure (2) between the Z-valued output of the
model Z yM,k , k 1,..., K and the observed real Z-valued data Z yk , k 1,..., K . The obtained result is
388 Fahreddin Sadikoglu et al. / Procedia Computer Science 102 (2016) 385 – 389

1 K
S ( Z yM , Z y ) ¦ Jk
Kk1
0.55 .

At the fourth step we use the Newton method based optimization algorithm to maximize overall criterion (2). By
applying the Newton method we obtained optimal value of the Jaccard similarity criterion as

1 K
S ( Z yM , Z y ) ¦ Jk
Kk1
0.63 .

The corresponding optimal values of the coefficients are a1 0.33, a2 0.33, a3 0.33 For illustration purposes
.
let us show the value of the model for the first row data Z yM,k 1 under the obtaned optimal coefficients (Fig. 1).
,

1 1
0,8 0,8
0,6 0,6
0,4 0,4
0,2 0,2
0 0
0,1 0,3 0,5 0,2 0,4 0,6 0,8

(a) (b)
M
Fig. 1. The Z-valued output of the constructed model Z y,k 1 ( A1 , B1 ) : (a) A1 (b) B1

5. Conclusion

In this paper we construct a Z-number based regression model for analytical description of dependence of
educational achievement on psychological determinants such as motivation, attention, and anxiety on of students.
The construction of such model is conditioned by the fact that observed data related to the considered phenomenon
are perception based and are characterized by fuzziness and partial reliability. The given data are described by Z-
numbers. The adequacy of the constructed model is formally justified by using the Jaccard similarity measure which
shows validity of the suggested approach.

References

1. Adedeji T. The impact of motivation on student’s academic achievement and learning outcomes in mathematics among secondary school
students in Nigeria. Eurasia Journal of Mathematics, Science & Technology Education 2007;3(2):149-156.
2. Aladro-Gonzalvo AR., Machado-Díaz M, Moncada-Jiménez J, Hernández-Elizondo J, Araya-Vargas G. The effect of Pilates exercises on
body composition: A systematic review. Journal of Bodywork & Movement Therapies 2012;16:109-114.
3. Caldwell K, Harrison M, Adams M, Triplett NT. (). Effect of Pilates and taiji quan training on self-efficacy, sleep quality, mood, and
physical performance of college students. Journal of Bodywork and Movement Therapies 2009; 13(2):155-163
4. Chapell MS, Blanding ZB, Silverstein ME, Takahashi M, Newman B, Gubi A. Test anxiety and academic performance in undergraduate
and graduate students. Journal of Educational Psychology 2005;97(2):268-274.
5. Ungaro A. Pilates: Body in Motion. London: Dorling Kindersley Publishing, Inc; 2002.
6. Bardossy A. Note on fuzzy regression. Fuzzy Sets Syst 1990; 37:65–75.
7. Diamond Ph. Fuzzy least squares, Inform Sciences 1988;46:141–157
8. Kacprzyk J, Fedrizzi M. (eds). Fuzzy Regression Analysis. Warsaw: Omnitech Press, 1992.
9. Meyer K. Random regression analyses using B-splines to model growth of Australian Angus cattle. Genet. Sel. Evol. 2005;37: 473–500.
 Fahreddin Sadikoglu et al. / Procedia Computer Science 102 (2016) 385 – 389 389

10. Angela BF, Ana C, González-Rodríguez G. Linear regression analysis for interval-valued data based on set arithmetic: a review. towards
advanced data analysis by combining soft computing and statistics. In: Christian Borgelt, María Ángeles Gil, João M.C. Sousa, Michel
Verleysen, editors. Towards Advanced Data Analysis by Combining Soft Computing and Statistics, Berlin: Springer; 2013; p. 19-31.
11. Zadeh L. The Information principle. In Proc. the 15th IEEE International Conference on Information Reuse and Integration, San Francisco,
CA, USA; August 13-15, 2014.
12. Zadeh LA. A note on Z-numbers. Inform. Sciences 2010;181:2923–2932.
13. Yager RR. On a view of Zadeh's Z-numbers. Advances in Computational Intelligence, Communications in Computer and Information
Science 2012;299:90-101.
14. Yager RR. On Z-valuations using Zadeh`s Z-numbers. International Journal of Intelligent Systems 2012;27:259-278.
15. Zadeh LA. Methods and systems for applications with Z-numbers. United States Patent. Patent No.: US 8,311,973 B1, Date of Patent: Nov.
13, 2012.
16. Aliev RA., Memmedova K. Application of Z-Number based modeling in psychological research. Comput Intell Neurosci. 2015; 2015: 7
pages (Article ID 760403). doi: 10.1155/2015/760403.
17. Sadikoglu F, Huseynov OH, Mammadova K. Modeling the effect of pilates exercises by Z-regression. In: Proc. Eighth International
Conference on Soft Computing, Computing with Words and Perceptions in System Analysis, Decision and Control, Antalya, Turkey,
September 3-4, 2015, p. 69-80.
18. Memmedova K. Structural modeling impact of Pilates on motivation, attention, anxiety & educational achievement. Procedia Social and
Behavioral Sciences 5th World Conference on Learning, Teaching and Educational Leadership WCLTA-2014, Prague, Czech Republic:
Volume 186, 13 May 2015, p. 544-548 Elsevier.
19. Ramli N, Mohamad D. On the Jaccard index with degree of optimism in ranking fuzzy numbers. In Proc. 13th International Conference on
Information Processing and Management of Uncertainty in Knowledge-Based Systems. Applications. Dortmund, Germany, 2010; p. 383-
391.
20. Aliev RA, Alizadeh AV, Huseynov OH. The arithmetic of discrete Z-numbers. Inform Sciences 2015;290: 134-155
21. Zadeh LA. Probability measures of fuzzy events. Journal of Mathematical Analysis and Applications 1968; 23(2):421–427.