Test case optimization and prioritization using data mining
techniques – An empirical studies
Shivani singh
Computer science and engineering, Krishna Institute Of Engineering and Technology
Ghaziabad - Meerut Highway (NH-58) - (AFFILATED TO UPTU) - INDIA
Email – shoisingh@gmail.com
ABSTRACT
Software engineering is a major process in SDLC
(SOFTWARE DEVELOPMENT LIFE CYCLE).Where
testing is a major area in this which consists of a set of
activities conducted with the aim of finding errors in software.
It is also a application of a systematic, disciplined,
quantifiable approach to the development, operation, and
maintenance of software, .In this paper we review on all the
previous worked based on data mining techniques based on
test case optimization and prioritization including models
and algorithm which give most accurate output by minimizing
the error .
Keywords
Test Case Prioritization, test generation , test cases, UML, data
mining, tabu search algorithm, Test Case Optimization,
genetic algorithm.
I.
INTRODUCTION
Nowadays software engineering play a vital role in
developing project, because most of the companies spend
almost 50 % amount of budget to this area. Software
engineering is quality oriented techniques where Quality can
be achieved by very good testing techniques.
Figure 1:- Software Testing module
Where software engineering consist set of instructions and
programs, each programs have their own functionality
depending up on its type. while Software Testing consists of a
set of activities conducted which finding errors in software.
Testing can b done by either manually or automatically.
Testing automatically is best way to test software
because it consume less time and give accurate result
than manual testing. Test case can also generate from the
UML using object diagrams. UML is unified modeling
language which is used to create visual models of a software
system.. We used dynamic technique of ―Tabu‖ search
method [1] for test case generation called as TCGen(test case
generator) which generates the most suitable test cases and in
the ground work necessary for test case prioritization. It has
been estimated the 50 percent of software development [2],
thus, a subset of all possible test cases has to be determined
that satisfies a particular criteria. As per for future aspect,
modification is also performed in the software depending up
on the request that is given by the user. The user can give the
change request form depending on their need. i.e modification
performed in their organization or depending upon the poor
performance of the software . Quality proportional to the
insurance of good testing techniques. So optimizing time and
cost of testing process are the real challenge of a test
engineers. Where Regression testing is a
kind of testing which requires maximum effort, time and
cost. Previous work on test case prioritization is
based on the computation of a prioritization index, which
Determines the ordering of the test cases (e.g., by decreasing
values of the index). For example, the coverage level
achieved by each test case was used as a prioritization index
[3]. Another example is a fault proneness index computed
from a set of software metrics for the functions exercised by
each test case [4]. Among several approaches that are used for
the testing, we are discuss specification-oriented approaches
(or black-box testing) and implementation oriented approach (
or white box testing ). Black box testing which generate the
test cases from the program specification, and white-box
testing which generate the test cases from the code of the
program under test. We have two different type of model of
prioritization i.e system base prioritization method and model
based prioritization. where system based modeling is used to
capture some aspects of system behavior or it is state based
for example real time system. And in model-based test
prioritization [5,6] a system’s model(s) is used to prioritize
tests. Futher it will divide into one type and second type
model , one type model is appropriate for prioritization based
modification that involve changes in the model and then in
the source code. While in another type of model-based test
prioritization methods [7] are appropriate for modifications
that do not involve any changes in models (changes are only
made in the source code). In this paper, we are discuss UML
2.0 for modelling which is based on the second type of
Model-based prioritization method. We are also brief discuss
about clustering technique[8] using genetic algorithm method
of data mining to prioritize the test case by tabu search
[9].Our major goal is to minimize time taken by test cases .
II.
RELATED WORK
This section include all the previous work based on test cases
optimization techniques based on proposed data mining and
software testing models , regression techniques using UML
diagrams ,genetic algorithm , object oriented prioritization
various approaches etc. to gain optimal result . These research
works contain their own advantages and disadvantages.
Alessandra Cavarra, Thierry Jeron, Alan Hartman ISSTA
(2002) in this paper researcher proposed a architecture for
model-based testing using unified modeling language (UML).
In which Class, object and state diagrams are used to define
model. To generate the test cases automatically, different
types of tool are available like AGEDIS (Automated
Generation and Execution of Test Suites in Distributed
Component-based Software) test generation tool. The main
advantage of the AGEDIS test case generation tool is its
ability to combine different test directives: coverage criteria,
test purposes and test constraints. This allows the user to tune
the selection of test cases with respect to the budget of the
campaign [10]. A hierarchy of test suites can be constructed
with the help of property that the larger the suite, the greater
the coverage of the implementation. This hierarchy is
particular useful in regression testing.
L.C.Briand el al. (2008) in this paper researcher proposed,
method supported by the prototype tool to deals with the
design level regression test selection problem by avoid human
errors. The main objective is to ensure that regression testing
was safe while minimizing regression efforts. But they show
that certain changes may not be visible in the design and may
require additional attention during coding or special way to
document them design. Another disadvantage is that, based on
UML design information, test selection may not be a precise
as if it was based on detailed code analysis. The case study
that is considered has only one case of imprecision in
classifying a test case as retesting [11].
M.Prasanna el al. (2009) in this paper researcher proposed
that to testing a software, test cases generation is best way.
The test cases are derived by analyzing the dynamic behavior
of the objects due to internal and external stimuli [12]. Test
cases can be generated with the help of UML diagrams.
Researchers use model based approach in which genetic
algorithms crossover technique is apply on the class diagram
and the traversal is done by the depth first search(DFS)
algorithm. This tree structure approach coupled with genetic
algorithm shows that it is capable to reveal 80% faults in unit
level and 88% faults in integration level. They couple the
genetic algorithm with mutation testing to check the
effectiveness in the testing process which shows 80.3% of
effectiveness. The result shows methodology is useful to
generate test cases after the completion of the design phase
and error could be detected at an early stage in software
development life cycle.
Sequence
Diagram
General
Tree
Genetic algorithm
Tree structure
s
Testing equence
Testing implementation
Testing
condition
Figure 2 :- flow diagram of genetic algorithm
A. Mosavi (2010) in this paper researcher proposed the
classification task of data mining has been introduced
as an effective option for identifying the most effective
variables of the MOO in MCDM (Multicriteria decisionmaking )systems. The Classification algorithm of LADTree
was utilized analyzing the effect of each design variable to the
indentified objectives. The number of the optimization
variables has been managed very effectively and reduced
in the given example.
The modified methodology is demonstrated successfully
in the framework. The author believes that the process is
simple and fast. Variables were reduced and organized
utilizing classification algorithms. The achieved
preprocessing results as reduced variables will speed up
the process of optimization due to delivered smaller design
space and minimum requested computational cost for
MOO process. Data mining tools have been found to be
effective in this regard. It is evident that the growing
complexity of MCDM systems could be handled by a
preprocessing step utilizing data mining classification
tools.
Sangeeta sabhwal, Ritu sibal, Chayanika Sharma (2011) in
this paper researcher proposed another novel based approach
in which testing efficiency is optimized by applying the
genetic algorithm on the test data. For requirement change, a
stack based approach for assigning weights to the nodes
object diagram is proposed [8]. Here the object diagram that is
considers are activity and state chart diagram. In this they
convert first activity diagram into CFG and state chart
diagram into SDG and then stack based approach is applied.
The test paths are generating from the activity diagram and
state chart diagram. In this novel approach the research uses
the genetic algorithm which is apply on the sequence diagram.
In the first sequence diagram is generated and then from the
sequence diagram. Sequence dependency graph is generated
and genetic algorithm is apply on it .In their study the found
that the approach used is significant to identify location of the
fault in the implementation and thus reduce the testing efforts
[13]. The proposed approach makes use of IF model and
genetic algorithm to find the path to be tested first.
A.V.K Shanth et al. (2011) in this paper researcher proposed
another model based approach in which the concept of data
mining is used in which the evolutionary genetic algorithm
technique is apply on the class diagram and generate the test
cases [12]. Researcher study that Data mining technique is
implemented to generate the optimal test cases automatically
by which human and cost efforts are minimized [12]. They
show that evolutionary genetic algorithm yields optimal valid
test case than with only genetic crossover operator, after
applying depth first searching algorithm. The advantages are
that specification –based testing uses.
G.Mohan Kumar, A.V.K.Shanthi (2012) in this paper
researcher proposed researchers used some novel approaches
to test the software at the initial stage itself which will easy
for the software testers to test the software in the later stage
[12]. In their study they use the tabu search algorithm to
generate the test cases automatically from the object diagram.
Here they take the sequence diagram. The experiment results
show that this method has better performance. All the possible
test cases are generated and validated by prioritization. test
cases which are generated can be used as test suite for path
testing for application. This approach can reveal all path’s for
software to be developed and also obtained test cases valid
once, which avoids validation test case because of fitness
criteria.
Ranjit Swain, Vikas Panthi, Durga Prasad (2012) in this
paper researcher proposed different techniques are used to
generate the test cases to test the software. The functional
minimization technique is also used to generate the test cases.
In this technique in which the STUPEC [13] technique is used
in which first predicate is selected and then predicated is
transformed and then test cases are generated. The functional
minimized technique is used for finding the minimum of
predicate function. In this approach the test cases are
generated step by step. Here the object diagram that is used
for generating the test cases is state machine diagram. This
approach covers much coverage like state coverage, transition
pair coverage, action coverage. The numbers of test cases are
minimized that achieve transition path coverage by testing the
borders determine by simple prediction. It is found that test
cases are generated from the object diagram by minimize the
cost and time [13]. It can also handle transitions with guards
and achieves transition path coverage.
Chien-Li Shen* and Eldon Y. Li (2013) in this paper
researcher proposed a research model by utilizing the Ant
Colony Algorithm and test case prioritization technique to
optimize the test case prioritization. The expected contribution
is to help software quality assurance stakeholders identify the
higher-risk test cases which enable project managers to
control testing execution time and testing budget. To validate
the result of prioritization in this model, the result will
compare with the one of a set of test cases chosen by
experienced test leaders in each run of regression test along
with the APFD (Average. Percentage Faults Detected) value.
In addition, the parameters in the algorithm will be adjusted in
order to dig out the best suitable variable values in this
algorithm.
V.Mary Sumalatha, G.S.V.P.Raju (2013) in this paper
researcher proposed the test case generation by means of
UML sequence diagram using Genetic Algorithm from which
best test cases can be optimized. Moreover this method for
test case generation inspires the developers to improve the
design quality and to find multiple test cases ready for
execution. In future, it is possible to build an automatic tool
using this approach. This automatic tool will reduce cost of
software development and improve quality of the software
A.Pravin and Dr. S. Srinivasan (2013) in these
paper researchers proposed The proposed Algorithm
introduces test case prioritization technique with the help of
the value calculated. The proposed algorithm works for both
requirement and testing. [15]The Experiment is done for
projects. Test case values were being used to rate the test
cases. Algorithm was compared with random prioritization
technique on two Application projects and it describes the
effectiveness of algorithm for early rate of fault detection. We
are currently working to see the effect of proposed algorithm
with other techniques. Additionally the proposed algorithm is
tested on limited data set. It can be validated by taking large
size projects having larger set of use cases and test cases. so
that it will be suitable for all type of projects with larger
complexity and length.
III.
CONCLUSION
In this paper we see different types of test case
approaches, methods, techniques of data mining used
to generate test cases after the completion of the
design phase and errors can be detected at an early
stage in the software development life cycle. or
optimization and prioritization of test case. Which
minimizing cost , time , effort , errors ,delay and
maximize accuracy and relevancy .and gives optimal
result by automatic generating test cases.
REFERENCES
[1] Bertolino.A, “Software Testing: Guide to the software engineering body of knowledge”, IEEE Software, Vol. 16, 1999, pp.
35-44.
[2] Clay E.Williams, “Software Testing from UML Specifications”, Second International Conference on Hitesh Tahbildar,
Bichitra Kalita, “Automated Software Test Data Generation: Direction Of Research”, International Journal of Computer Science
and Engineering Survey (IJCSES) vol.2, No.1, Feb 2011, pp. 99-120.
[3] Alessandra Cavarra, Thierry Jeron, Alan Hartman, “Using UML for Test Generation”, ISSTA, 2002.
[4] M. Prasanna, K.R.Chandran, “Automatic Test Case Generation for UML Object Diagrams Using Genetic Algorithm”, Int. J.
Advance. Soft comput. Appl., vol.1, no. 1, July 2009, pp. 19-32.
[5] L.C.Briand, Y. Labiche, S.He, “Automating Regression Test Selection based on UML designs”, Information and Software
Technology 51(2009)16-30.
[6] A.V.K.Shanthi, Dr.G.Mohan Kumar, “Automated Test Cases Generation for Object Oriented Software”, Indian Journal of
Computer Science and Engineering (IJCSE), vol.2, no.4, Aug-Sep 2011, pp. 543-546.
[7] Sangeeta Sabharwal, Ritu Sibal, Chayanika Sharma, “ Applying Genetic Algorithm for Prioritization of Test Case Scenarios
Derived from UML Diagrams”, International Journal of Computer Science Issues, vol.8, issue 3, no.2, May 2011, pp. 433-444.
[8] Marlon Vieira, Johanne Leduc, Bill Hasling, Rajesh Subramanyan , Juergen Kazmeier, “ Automation of GUI Testing Using a
Model – Driver Approach”, Proceeding of International Workshop on Automation of Software Test, 2006, pp. 9-14.
[9] Manoj Kumar, Dr. Mohammad Husain, Gyanendra K. Gupta and Amarjeet Singh, “An Efficient Algorithm for Evaluated for
Object Oriented Models”, International Journal of Computer Application, vol.24, no.8, 2011, pp. 11-15.
[10] Ranjit Swain, Vikas Panthi, Prafulla Kumar Behera, Durga Prasad Mahapatra, “Test Case Generation Based on State
Machine Diagram”, International Journal of Computer Information Systems, vol.4, no.2, 2012, pp. 99-124
[11] A.V.K.Shathi, G.Mohan Kumar, “ A Heuristic Approach for Automated Test Case Generation From Sequence Diagram
Using Tabu Search Algorithm”, European Journal of Scientific Research, vol.85, no.4, Sep 2012,pp. 534-540.
[12] A.V.K.Shanthi, G.Mohan Kumar, “ Automated Test Cases Generation from UML Sequence Diagram”, International
Conference on Software and Computer Application, vol.41, 2012, pp. 83-89.
[13] Philp Samuel, R.Mall, A.K.Bothra, “Automatic Test Case Generation Using UML State Diagram”, IET Software, 2008, pp.
79-93.
[14] Ashalatha Nayak, Debasis Samanta, “Automated Test Data Synthesis Using UML Sequence Diagram”, Journal of Object
Technology , vol. 9,no.2, 2010,pp. 75-104.
[15] A.Pravin “An Efficient Algorithm for Reducing the Test Cases which is Used for Performing Regression Testing “2nd
International Conference on Computational Techniques and Artificial Intelligence (ICCTAI'2013) March 17-18, 2013 Dubai
(UAE)