Abstract systems. A little of the reimbursement

0 Comment

— Software defect
predicting was a challenge all the time till now. Unfortunately we are still
blind to know that when we can be able to ask that a new developed software is
the perfect product for the required business or industry and no more changes
and developments required. Many companies want to predict the software defect
before they bring it in production mode to help then to save the time and
money. In this paper we will discuss about the different technique and software.

Keywords — Software
faults and failures;Support vector machines; formal methods.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now


With the time software systems
become more complex and we do not know that what will the results over the
time. Every algorithm has different behaviors on different data. Some time an algorithm
is working very efficiently upon a small data and as its data increased the
results become totally reversed. On the other hand we can say that each day we
have a new challenge regarding software. We are still blind at what will the
result on over next query. In prospect real-time software systems should be robotically
adapted based on the requirements for specific implementation situation and
working conditions It contains dynamic code synthesis that makes modules to
provide the functionality necessary to carry out the preferred process in

However, it requires a real-time review technique that must
be incorrectly incorrect / error free, since they classify dynamically
generated systems. A little of the reimbursement of dynamic reliability judgment
take in providing feedback to the operator to change the mission-objectives if
the dependability is low, the opportunity of masking defects at run time, and
the possibility of practical reliability management. One loom
in achieving this is to use software defect prediction techniques that can
assess the reliability of these systems using defect metrics that can be
dynamically measured. A
variety of software defect forecast techniques have been projected, but none
has established to be consistently exact

. These techniques include
statistical methods, machine learning methods, parametric models, and mixed
algorithms. Clearly, there is a need to find the best forecast technique for a
given prediction problem in context, and, possibly, bring to a close that this
problem is largely impenetrable. The availability of a large number of
divergent prediction techniques poses some interesting questions that need
great attention to details. Some of

these are, 1) What
characteristics of the underlying defect data influence the prediction accuracy?
2) Can we systematically determine the relationship between the accuracy,
choice of prediction system, and the data set characteristics? 3) How many
times do we need to repeat the sampling and validation procedures before
acquiring the required confidence intervals on the model? 4) Can we combine
different techniques for better prediction? If so, what are the trade-offs? 5)
Do software defect data have some characteristics (such as, standard deviation)
in common?  This paper provides a
critical review of software defect prediction techniques with special emphasis
on machine learning based methods. These techniques are applied to three
real-time defect data sets obtained from NASA’s MDP (Metrics Data Program) data
repository. Section 2 explains the probable characteristics of the defect data
that influence the accuracy of the prediction techniques. Section 3 reviews
some of the related works in this field. Section 4 details the software used
and defines the contents of the data sets used for the experiment. Section 5
explains the different prediction techniques and their application to the above
data sets. Section 6 compares the relative performances of the prediction
techniques under scrutiny. Section 7 concludes by giving some future


I'm Rick!

Would you like to get a custom essay? How about receiving a customized one?

Check it out