Scribd Logo
Upload

Welcome to Scribd!

  • 23 views

    Reducing Features To Improve Bug Prediction.: Exercise 1

    Uploaded by

    nijar
    This paper proposes a technique to reduce features to improve bug prediction in software. The technique applies feature selection to classification-based bug prediction, which addresses issues with existing classifiers - insufficient scalability and accuracy when using a large number of features. The paper clearly presents the problem, designed artifact, design process, and evaluation through empirical analysis on multiple projects. It contributes by showing which classifier choices lead to best bug prediction performance using feature selection.

    Copyright:

    © All Rights Reserved

    Reducing Features To Improve Bug Prediction.: Exercise 1

    Uploaded by

    nijar
    23 views3 pages
    This paper proposes a technique to reduce features to improve bug prediction in software. The technique applies feature selection to classification-based bug prediction, which addresses issues with existing classifiers - insufficient scalability and accuracy when using a large number of features. The paper clearly presents the problem, designed artifact, design process, and evaluation through empirical analysis on multiple projects. It contributes by showing which classifier choices lead to best bug prediction performance using feature selection.

    Original Description:

    hlw man

    Original Title

    Exercise 1

    Copyright

    © © All Rights Reserved

    Did you find this document useful?

    Is this content inappropriate?

    This paper proposes a technique to reduce features to improve bug prediction in software. The technique applies feature selection to classification-based bug prediction, which addresses issues with existing classifiers - insufficient scalability and accuracy when using a large number of features. The paper clearly presents the problem, designed artifact, design process, and evaluation through empirical analysis on multiple projects. It contributes by showing which classifier choices lead to best bug prediction performance using feature selection.

    Copyright:

    © All Rights Reserved
    23 views3 pages

    Reducing Features To Improve Bug Prediction.: Exercise 1

    Uploaded by

    nijar
    This paper proposes a technique to reduce features to improve bug prediction in software. The technique applies feature selection to classification-based bug prediction, which addresses issues with existing classifiers - insufficient scalability and accuracy when using a large number of features. The paper clearly presents the problem, designed artifact, design process, and evaluation through empirical analysis on multiple projects. It contributes by showing which classifier choices lead to best bug prediction performance using feature selection.

    Copyright:

    © All Rights Reserved
    You are on page 1of 3

    Exercise 1

    This work is performed by an individual


    Title of the paper

    Reducing Features to Improve Bug Prediction.


    Objective of theresearch
    1. This paper actually based on software bug prediction by using system log data.
    2. Additionally, this paper also figured out two drawbacks of existing classifier based on bug
    prediction which are potentially insufficient for practical used for large number of features. In this
    regard, this large number of features has huge impact on scalability and accuracy of the approach.
    3. The main objective of this paper is to proposing a technique which is a feature selection technique
    applicable to classification-based bug prediction.
    Problem relevance (is it presented and how?)
    1.The main problem that has described within the paper is that existing classifier based on bug
    prediction of the software log data. When we are using large number of features it causes the
    problem of large number scalability which will leads to potentially inaccuracy for the prediction
    of the bugs.
    2.The other relevance problem is that which classifier is best for the software bug data and what
    metric should be best approach for predicted the bug successfully.
    Designed artefact (what it is and is it clearly presented?)
    For Design artifact concept, at first the authors are tried to identify the problem from the existing
    classifier problem about the software bug prediction. Within the article authors are discussed with their
    research objectives and solutions for the eradicating this problem. For the accuracy identification they
    were used features selection technique and later for the validation of the features method they are using
    confusion matrix for clarify the solution of the research. From the design research they actually design a
    new artifact based on existing solution and make some changes based on the prediction which will lead
    their research to make the solution. This article has clearly described all those limitations of the existing
    classifier problem and also predicting the software bug using their own design artifact successfully.
    Design process (is it described and how design was done?)
    From the very beginning of the article the authors are tried to focus the problem and also maintain their
    own design artifact that has hugely used whole research area of the research for predicting the software
    bugs. For the demonstration phase of DSRM model, they are used by their own design artifact for come
    up a solution. For using large number of features and also having the limitation of existing classifier they
    were proposed a design artifact which was actually feature based technique which will reducing the
    features for the software bug prediction. For clarify the design artifact that they proposed through their
    technique is used for empirical analysis that was actually found the result analysis for the evaluation the
    result. The article has well described and also come up a new design artifact of technique which will later
    prove the solution of the current problem.
    Design evaluation (was evaluation done and how, was it presented clearly?)

    Design evaluation has done through the empirical analysis of the solution’s result. In that paper
    the two main variables were main affecting the bug prediction performance and they are (1) type
    of classifier (Naïve Bayes, Support Vector Machine), and (2) which metric (accuracy, F-
    measure) the classifier is optimized on. They were using the four permutations of these variables
    are explored across all 11 projects in their data set. For each project they have performed metric
    calculations like accuracy, buggy precision, buggy recall and buggy F-measures. From DSRM
    model they were observed their result and tabulated their data in to their research paper so that
    they can observe and make the decision about their artifacts. In later they have mentioned the
    effect of their feature’s technique based on their observations. Finally, they have successfully
    established their design artifact which will leads the solution for the current problem regarding
    the software bug prediction.

    Research contributions (what were the contributions and were they clearly presented?)

    The main contribution that has come up from an answering some research questions whose
    are found within the article. At first, they tried to show that which choices lead to best bug
    prediction based on the performance using feature selection. Secondly their main contribution is
    that I noticed about the range of performance of the best-performing Bayesian (F-measure
    optimized) classifier across all projects when using feature selection.

    Research rigor
    The research rigor of this article is Quantitative based on the empirical analyses from different projects.
    At first the author wants to identify the problem and then showed their research design artifact of
    technique for reducing the problem. Later they have analyzed empirically from different projects of data
    and evaluate their design through their prediction of analysis.

    Was it easy to find the issues in the paper?


    Yes, this paper has demonstrated all those approaches that has fulfilled based on DSRM model and also
    their analytical result satisfied their prediction through their design artifact.
    Was the paper good, why / why not?
    This paper is good because this paper follows a systematically approach based on DSR approaches. At
    first their problem identification they have showed and also their design artifact based on the solution of
    the problem. Later, they will follow the empirical analysis for the evaluated their design artifact. So this
    paper actually good and also clearly well described and figure out lots of interesting regarding the
    software prediction of bugs.
    References:

    Shivaji, S., Whitehead Jr, E. J., Akella, R., & Kim, S. (2009, November). Reducing features to
    improve bug prediction. In 2009 IEEE/ACM International Conference on Automated Software
    Engineering (pp. 600-604). IEEE.

    You might also like

    pFad - Phonifier reborn

    Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

    Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


    Alternative Proxies:

    Alternative Proxy

    pFad Proxy

    pFad v3 Proxy

    pFad v4 Proxy